You may have noticed I left something out in that last post about Andreesen and Horowitz — their political vision is focused on crypto, AI, and a tax policy they like better
. I said nothing about AI! You may praise me for my restraint.
I shall now correct my omission. Here’s the breakdown of the money OpenAI is spending on this boondoggle:
Total revenue has been $283 million per month, or $3.5 to $4.5 billion a year. This would leave a $5 billion shortfall.
Training the AI models will cost OpenAI about $7 billion in 2024. For ChatGPT alone, the cost will be $4 billion. New models may add $3 billion to that cost — OpenAI has had to train new AI models faster than it had anticipated.
Microsoft’s OpenAI “funding” is largely in the form of Azure compute credits. OpenAI gets a heavily discounted rate of $1.30 per A100 server per hour. OpenAI has 350,000 such servers, with 290,000 of those used just for ChatGPT. This cloud estate is apparently running near full capacity.
Staffing costs for 2024 are likely to be $1.5 billion, up from $500 million for 2023. The median OpenAI engineer salary in 2023 was a $300,000 base salary and $625,000 of stock-equivalent compensation. [Bay Area Inno, 2023; Levels.fyi]
They’re spending roughly twice what they’re making. Almost all their servers are chewing away on ChatGPT. And personally, the worst of all as far as I’m concerned is that software engineers are getting paid $300,000 base salary and $625,000 of stock-equivalent compensation. If my employer paid me half that amount of base salary, rather than a quarter, and never mind the big stock bonus, I’d be coasting on easy street and could hire a live-in masseuse and, I don’t know, go crazy and buy a second car? I struggle to imagine that much money.
I suppose my daughter would have the qualifications to get into that kind of business, but I’d encourage her to keep her soul intact.
Recent podcast episodes:
Better Offline – “Pop Culture”:
Tech Won’t Save Us – “Generative AI is a Climate Disaster”:
Oh look, another dot-com bubble exploding? A few techbros sweating bullets and realizing what idiots they were, yet again? Color me surprised, not at all.
I suppose that after this nth hype-cycle has passed, we’ll see what the real, useful applications of AI are going to be but first it seems we need a general financial crash. Yet again.
What was that about humans being the rational animal? Hhh, call me grumpy.
When ChatGPT came out, I thought it was a neat toy. It was fun to explore what it could come up with as I bounced weird sci-fi and fantasy ideas off the stochastic parrot. But seriously, techbros, it’s time to give it a rest and explore new angles if you want something that mimics intelligence. The stochastic parrot has run its course. Cue the Monty Python dead parrot sketch.
For a lot of reasons, OpenAI and LLMs in general may be doomed, but I’m fairly sure that the way to look at those numbers is as an investment in future profits…in other words a bet, a gamble…and not the immediate cost versus benefit (income). If it works, OpenAI, Microsoft, Google, FaceBook, Apple, and all the other players stand to make a lot of money from it. If it doesn’t pan out, then it was a bad investment they can write off. The entire digital tech revolution of the last 50 years or so is based on that sort of approach.
Now, should society be investing in education at that kind of scale, including paying seasoned professional educators at every level that kind of salary? Absolutely. A well educated population is one of society’s most secure investments even if the immediate return is equally poor. In my not-so-humble opinion, If we can afford to invest those kind of dollars for a shot in the dark technology like genAI and LLMs, then we can afford to invest those kind of dollars to better educate people.
As best I can tell, AI is still a failure. The “artificial” part works but the “intelligence” part doesn’t’
Yes, they are using clever tricks to make it look intelligent. But the intelligence comes from the humans. The LLMs (large language models) are just plagiarizing little bits of human intelligence and trying to synthesize those into a system that gives a superficial appearance of being intelligent.
There do not appear to be any real breakthroughs in understanding intelligence.
robro:
The reason those companies got so big is that very sort of investment; they are investing to make money from it, not to help people in general.
They will of course invest in their human assets, to make them more productive.
Anyway, they can afford it, but we cannot, because they have those kind of dollars and we don’t.
I remember something I saw a few weeks ago about AI and how it’s a fraud. It pointed out that when genuinely revolutionary new pieces of tech come out, it’s easy to see what they can be used for. The iPhone, for example. With AI, all we’re seeing is how to use it to commit plagiarism and manufacture deepfakes.
Not everyone shares your opinion, microraptor.
e.g.
https://www.leewayhertz.com/ai-use-cases-and-applications/
or
https://www.forbes.com/sites/qai/2023/01/06/applications-of-artificial-intelligence/
etc.
(Profit is the motive)
In full climate crisis, a few techbros have decided to ramp up our CO2-equivalent so that they can better answer the whims of their fragile egos and rake in VC money.
These people have enough brain to understand the problem, they just don’t give a fuck.
And half of humanity worships them.
@John, #8
You’re the last guy around here I’d have expected to uncritically swallow the excretions of the exited capital-class. Can you honestly not see the hype bubble at work here? Let’s look at your examples of AI ‘utility’ that you offered. Taking the first link you provided:
Well, I’m completely convinced. It’s really great that AI is going to help businesses integrate their mulitfacted pain points and propel them to unprecedented levels of excellence. That’s clearly a substantial couple of paragraphs that summarize coherent explanations about what the technology is actually going to do, and definitely not a remarkably complete collection of corporate-speak buzzwords strung together in an almost-compehensible fashion.
Promising start. With this as an appetizer, I can’t wait for the main course. Tell me, o oracle, what magic the god-in-a-box will do to improve my life, or at least to improve shareholder returns.
My god. I’m overwhelmed. I love tailored use-cases. Not only will the AI be a strategic ally, I hear that it’s going to reshape the traditional paradigms! And all that while we’re in an era of digital acceleration? Someone fetch the fainting couch, I’m completely overcome by the obvious utility of the underlying concept here. I must press on! What amazing insights are in store if I continue reading John Morales’ preferred explanations of the utility of AI software? I can’t miss this; if I do, my paradigms might remain in their current obsolete shape!
I’m not going to go on, the bit is probably tedious already. I can’t honestly imagine that John actually read that vapid article and thought it contained insights worth sharing. After we get through the tedious repetition of corporate bafflegab, it just descends into a list of things that generative AI could, maybe, conceivably, do – if you ignore or don’t care about it’s occasional tendency to outright invent things. An AI health advisor sounds like a good idea, if you ignore the documented history of telling people to eat rocks and put glue into their pizza sauce. They’ll probably fix that in the next iteration of GPT anyhow, and nobody really cares about when the engineers mention exponential training cost growth. What’s an exponent? Should we ask an AI? Is it a food? A new management strategy? I’m sure it’s important. Hopefully it can tailor my use-cases, because lately they haven’t been fitting in a flattering way.
AI is just a way for the pigeons of Tech industry leadership (pigeons always flock together – they might not always find food, but nobody wants to be the lone pigeon when a cat’s around) to signal to each other that they’re all on the same page. Sure, there’s no actual way that AI is going to make Travelodge more efficient or profitable, but if you’re the CEO, you have to say that you’re using or thinking about using AI, because everyone else is doing it, and you don’t want to be left out. Nobody wants to head into the future with an unreshaped paradigm, after all.
snarkhuntr, you must realise your expectations were… um, silly.
Duh. Of course I can. So what?
Point being, whether or not it’s being hype bubbled, it’s another step into automation.
I mean, come on!
Back in the day, could you imagine saying “paint me a picture in the style of so-and so with these garnishings” and have it appear?
That’s where people such as Neil Rickert lose the plot; it’s called AI, but it’s not AI.
It’s automation of stuff that people used to think required intelligence.
From mining to chemistry to animal identification to sorting peanuts.
All stuff that used to take an “intelligent” person to do, but now can be automated.
Unintelligent, sure. ;)
Random links, basically.
First hits on a Google search, which (like it or not) is probably AI-augmented.
Didn’t even read them, but be aware there are a shitload of them.
Point being, if billion-dollar companies are investing in it, they are doing it because they think $$$.
As are other enterprises.
Well, then.
You are not disputing me in the least, are you?
You have nothing, but you have it at length.
—
Anyway.
Those who have the money and the nous and wish to $$$ from it, they are doing AI.
cf. https://en.wikipedia.org/wiki/Nvidia
That ain’t hype, that is reality.
I remember AlphaZero playing chess, and then go, and then SC2.
Back in the day (what, a few years ago now?), and then it was only impossible to beat at chess; it won at go and sorta played professional-level SC2.
(That was one of the first AI engines, it learnt to play by playing)
—
Care to define to what you intend to refer by the term “AI”?
Because, it’s already replacing people in the workplace.
First, it was the weavers. And, oh yeah, John Henry.
Now, the world chess champion and other super-GMs all acknowledge chess engines are to them as they are to regular players. Fact of life.
Etc.
Well, one more:
I am somebody, and I care not one whit whether or not I head into the future with an unreshaped paradigm.
Claim falsified, therefore.
(Beware of making universal claims without sufficient basis)
You know, we are both on the internet.
(Was also called a bubble, back in the day)
So. <clickety-click>
https://portfolioslab.com/symbol/NVDA
Huh.
(Would’ve been nice to invest in it, no?)
They invest in Dutch tulips, or snap up mortgage-backed securities, or jump off a cliff, you jump with them? <smh>
… says the guy who just linked to two content-free marketroid brochures as “
” that something wasn’t “ “. <smh>LLM-style AI has proven itself good at doing exactly one thing (aside from parting fools from their money, which almost anything can be made to do, under the right circumstances), and that is writing (and painting) fiction. And even then it behooves the user to proofread and edit the output as it often comes out a little less than polished …
And John: what the heck has happened to you recently? I know you’re capable of doing better than this. Because you were just a week or so ago …
Bekenstein Bound, it is sad you are disappointed.
Bewildered, bemused, mazed!
Heh.
I am not a trillion-dollar enterprise, and the Dutch tulip thing is basically a myth.
Perhaps inform yourself.
cf. https://www.smithsonianmag.com/history/there-never-was-real-tulip-fever-180964915/
I thought a couple was sufficient, but there are hundreds of examples.
Thousands.
(It’s on you that your Google-fu is weak)
The topic is supposedly AI in general, not just LLM types of AI.
Mate!
I made my first comment on this blog (well, the first instantiation) back in November 2005.
I have always been me, it’s you that misperceived me and thus imagines that something has happened to me.
(But hey, at least you’ve moved on from claiming my comments are drunken ramblings)
PS
I dunno about you, but for me, when I ostensibly quote someone I quote them, I don’t paraphrase them on the basis of what I reckon they may have intended.
The actual term “
“ was typed by you, not by me.(Tsk)
Also, https://www.ibisworld.com/au/market-size/mortgages/
“The market size, measured by revenue, of the Mortgages industry was $101.2bn in 2023.”
(Why anyone would imagine that’s worthless is left to the imagination)
It was exaggerated for comedy and then the comedy was taken completely seriously by some historians, but the bubble did happen. It didn’t wreck the whole economy, but it did rock a few boats.
We also saw the modern version in crypto and NFTs. There’s no reason to assume speculative investors are smart or rational by default. All I see is the madness of crowds and random computation feats being labeled as “AI” to attract these frenzied investors.
Crpyto is just a huge fraud to obscenely boost billionaire wealth.
On top of all that, think about how the public is paying for the massive data transmission over the internet and the massive amounts of electricity these server farms use are threatening the grid and causing big increases for everyone in electricity cost. Also, the fact that crypto facilitates so many frauds in job offerings, etc.
Here is another dangerous thought. If this country becomes a ‘cashless’ society and there is another crowdstrike/microsoft crash. almost all the financial transactions (no buying food with a credit card) in the entire country halt.
@11 John Morales:
Alphazero did not play SC2. Their AI thing for SC2 was not a “zero” (was not trained via self play) and it’s supremacy was at doing mechanically super human micro stuff that humans cannot do and it was very weak at the strategic stuff. It also could see the entire map and it could not really beat top players despite DM desperately wanted to do so. In one notable game, the human player survived because even though the AI had an overwhelming ground army, it could not figure out that it needed to build just a few units that can shoot at air units.
lotharloo, yes, it was https://en.wikipedia.org/wiki/AlphaStar_(software) by the same enterprise (DeepMind).
Be aware that in mid-2019 they constrained it to having to scroll the map window and low APM, and still reached GM level.
(https://deepmind.google/discover/blog/alphastar-grandmaster-level-in-starcraft-ii-using-multi-agent-reinforcement-learning/)
—
Recursive Rabbit @18,
You’re not seeing the whole picture.
cf. https://www.weforum.org/agenda/2024/05/these-5-countries-are-leading-the-global-ai-race-heres-how-theyre-doing-it/
PS
You know the old adage?
“There are old pilots, and there are bold pilots, but there aren’t old bold pilots”.
Basically, stupid or irrational investment capital providers don’t last very long.
cf. https://www.investopedia.com/terms/v/venturecapital.asp
Skimmed the thing about AI diagnostics, and… it seems they’re overestimating what ChatGPT can do.
What theoretical basis is there for believing a stochastic parrot could aid in diagnostics?
What makes you think you and I are not stochastic parrots?
But, to give you a straight answer: the system can basically winnow through all the possibilities in seconds, then present that for the perusal of a trained diagnostician. A time-saver.
We know what we don’t know, at least approximately and much of the time. LLMs have a fatal weakness in that area. Therefore there is at least 1 thing (and probably a lot more) to human cognition beyond what is in contemporary LLMs and similar systems, and their predecessors.
What, echoing Rumsfeld?
Yeah: “There are unknown unknowns”
LLMs are obviously not sufficient, but perhaps they may be necessary.
Like you know, the intuitive side of thinking, as opposed to the intellectual side.
Who claimed otherwise? :)
BTW, BB, are you aware of https://en.wikipedia.org/wiki/Cyc ?
Do ya remember the fuzzy logic bubble, back when?
(Now, it’s just a component for various systems)
—
Again: don’t get hung up on the label (arty intellligence!), but see it as what it is — the achievement of hitherto-novel abilities by mere machines, which had been thought to require intelligence.
(The relevance of their game-playing prowess is proof-of-concept)
<
blockquote>What makes you think you and I are not stochastic parrots?
<
blockquote>
Because I don’t string words together based on what word is historically likely to follow next.
Also, I think you missed BB’s point: LLMs are perfectly content to make up plausible-sounding nonsense rather than acknowledge there are gaps in its knowledge. That’s the danger of hallucinations. If you ask it for an answer to a question, it doesn’t collect actual knowledge or do research, it makes text that looks like a semi-human answer.
Remember that lawyer who had ChatGPT hallucinate some nonexistent case files to argue his case in court?
Recursive Rabbit, I’m pretty sure you missed my point.
That is, that LLMs are but one technology being labelled as AI.
(You didn’t get the significance of my adduction of Cyc, did ya?)
Depends on the training dataset.
GPT, for example, stands for Generative Pre-trained Transformer .
One could train those things on purely curated data, do you not get that?
—
Anyway, I will quote myself, since you appear to have missed the significance:
“The topic is supposedly AI in general, not just LLM types of AI.”
and, well, also
“LLMs are obviously not sufficient, but perhaps they may be necessary.”
(Way ahead of you)
<snicker>
You too fall into the trap of thinking the current pre-release iterations, which are a proof-of-concept, and the freebies available to the public (yes, to generate hype) are all there is to this nascent tech.
This is the earliest of early days, the tech is nascent.
Basically, they are tools.
And only a tool would imagine that the use of a tool is the only use case for that tool.
So, yeah, that lawyer who had ChatGPT hallucinate some nonexistent case files was an idiot.
(Here’s another adage: blame the worker, not the tool)
@john,
That your mode of argument is to google a question and then, without even reading them, post a couple of the first results as evidence is pretty telling. Not sure why you’d admit to that, but at least you’re honest. Your AI boosterism is starting to make more sense.
And you fall into the trap of believing that the carefully curated investment-seeking ‘proofs of concept’ arent the very, very, best that those companies have to offer. You also assert that this tech is nascent. I assert that it might actually be mature. I think we’re quite rapidly reaching the end of the period of development where rapid gains are made, and that future improvements are going to be tiny, incremental and slow. The costs of those improvements are also rising in a literally exponential fashion. Hell, Sam Altman is out there theorizing that entirely new kinds of power generation are going to be required to produce the results he assures us
But since you seem to accord the finance industry an unearned degree of respect – perhaps you’d be willing to read this report by noted vampires Goldman Sachs. (H/T to the excellent Ed Zitron)
https://www.goldmansachs.com/intelligence/pages/gs-research/gen-ai-too-much-spend-too-little-benefit/report.pdf
While the current crop of AI applications have produced some cool party tricks, and have indeed automated the production of marketing bafflegab, scammy ad-filled product review sites and churn-level art used for Youtube thumbnails; I haven’t seen any evidence that it’s got much utility beyond that. I think generative visual AI is probably going to end up confined to novelty applications, problematic or illegal pornography production, and as a specialized tool used by graphic artists to quickly produce the parts of their images that nobody cares much about. Inpainting is neat. The main function of ChatGPT will be to produce text that you read for a minute or two while scrolling past ever-more-desperate Ads before you realize that you’ve been tricked by a shitty low-effort AI generated site and close the tab.
Meanwhile, the finance industry has determined that “AI” is the buzzword du jour, and that magically invoking it will produce temporary rises in stock price – so we can expect the flurry of absurd pronouncements about companies adding AI to their business models to continue until the next hype cycle spins up. Just like they all did with ‘metaverse’ and ‘blockchain’ during those respective cycles. The tech industry hasn’t really been doing a lot of innovation or meaningful improvements lately, and with the death of free money they’ve got to do something to preserve the illusion and fantasy of endless growth without limit.
No, it wasn’t “called bubble back in the day”. There was something called “Dot-com bubble” but nobody called internet itself a bubble. Why was it called a dot-com bubble (and only after the fact)? Because it grew out of proportions and burst, spectacularly. See, when internet was fairly new, people have recognized it as useful technology in wide array of activities (actually useful unlike what’s called AI today) and they got really excited about it, but a bit too excited (just like with this AI bubble today). You only had to slap words like “internet” or “network ready” on your company and have a website and investors would flock to you even if you didn’t have a business plan or a product. Or a viable business. Hint: Dot-com bubble burst once that fact became apparent.
There are tactics that are both stupid and irrational in the long run but work in short run. And are illegal but whether something is legal or not is important only as long as laws are actually enforced.
https://pluralistic.net/2023/05/19/fake-it-till-you-make-it/#millennial-lifestyle-subsidy
Odd thing to write when it is a well known fact that somebody doesn’t need to be smart to be evil or a thief. I guess that goes into “rational capital provider” group for you: Being able to strip-mine a business and loot it so hard it shuts down. Including things like emergency rooms and hospitals!
https://pluralistic.net/2022/11/17/the-doctor-will-fleece-you-now/#pe-in-full-effect
snarkhuntr:
<snicker>
That you cherry-pick directly, instead of by insinuation, is also telling.
Again… how many hundreds, thousands of examples do you want?
It’s there.
You can be as purblind as you care to be, but reality is.
Only if one is as dim as you clearly are.
I’m not advocating for anything, I’m noting what the reality is as contrasted to the insipid and foolish claims people are making.
Heh.
Not even slightly.
Look, this is pointless.
.Perhaps try to stop trying to tell me what I think, and actually dispute my actual claims;
I seem to, do I?
Shows how jaundiced your perception has become.
Bah.
You are so feeble.
Your basic argument is that [“AI” is the buzzword] is equivalent to [“AI” is worthless].
Weak as fuck.
—
Look, you can dispute my claims, but to try to niggle at me is almost as futile.
jo1storm, I am adumbrating a period of time.
(You are just being simplistic)
I used to work in an IT shop (mainframe, Fujitsu) just as
FFS, I lived it. At the time.
The https://en.wikipedia.org/wiki/Protocol_Wars
—
Anyway, since you’ve made the effort, let me give you a comparison.
You (and others here) imagine the internet is connected via a dial-up modem.
These days, well. A tad faster, no>
—
More meta:
Good grief!
Used to be that people thought one had to be intelligent to be evil or a thief.
That’s one of my points!
You can automate what used to require human intelligence.
(Consider the capability of automation as a function of time, perhaps)
@john
I do dispute your claims, as have many others in this topic. Tellingly, when your claims are disputed you resort to churlish name calling and insult, rarely if ever responding to the points that people actually make in disputing your assertions.
For example – you’ve argued by the imputed authority of the financial investment currently being poured into ‘AI’ that it must have intrinsic value or why would such sophisticated investors put their money there. (please note, since you seem prickly about this, that this is a paraphrase and not a direct quote). When offered evidence of other members of that class of people taking a different approach you simply ignore it. After all, you’re not much interested in reading – as you’ve demonstrated.
It is also distinctly odd to be accused of cherry picking for responding to your chosen evidence in support of your claims. If you had better evidence, I would have assumed you’d have provided it.
As I’ve read it (aside from the diversion about SC2), The sum total of your points here has been that there are thousands of articles that talk breathlessly about how wonderful ai will be at some point in the future.. It is hardly irrelevant to point out that it is currently used mainly as a fundraising/stock price manipulating tool and that the finance industry treats recitation of “AI” as somehow proof of future value. Perhaps that, rather than actual value is the reason why you can find thousands of articles about it with just a casual google.
If there’s one lesson that the financial markets have taken over the last 40 years or so of tech bubbles, it’s that you don’t need to make anything good – you just need to make something that people think is going to be good long enough for the founders and VCs to cash out. Somehow the bagholders just keep lining up to buy a piece of that bubble…. “Look ma! It keeps growing endlessly! Let’s get a some of it for ourselves….. hey, where’d it go?”
What does this have to do with you onerously comparing the whole of internet with dot-com bubble?
Nobody thought that but you.
“Buy a hospital, lock all the doors except for ER door, charge everyone who enters the hospital ER fee for passing through the ER” is Sunday morning cartoon villain level of evil. It doesn’t require a lot of intelligence to come up with that scheme. But it goes against your statement of
so you ignore it.
Nah.
So far, you’ve first insinuated, and now assert you’ve disputed them.
<smirk>
You don’t follow convos very well, do ya?
BB@14: “They invest in Dutch tulips, or snap up mortgage-backed securities, or jump off a cliff, you jump with them? ”
Me @17: “Also, https://www.ibisworld.com/au/market-size/mortgages/
“The market size, measured by revenue, of the Mortgages industry was $101.2bn in 2023.”
(Why anyone would imagine that’s worthless is left to the imagination)”
See, that’s got nothing to do with imputed authority that you imagine, and everything to do with the fact that those claims are all silly.
Dutch thing? Not a thing.
Mortgage thing? A thing (cf. 2008 subprimes) but yet still worth $101.2bn in 2023 in Oz in 2023, and very much not worthless thereby.
As for cliff-jumping, care to provide some examples?
—
See, this is your thing.
You are what I call a cargo-cult disputer.
Never actually quote or try to refute, always impute bullshit to me and insinuate without justification.
(How’s that working for ya?)
—
Hey, you’ve found your snark!
(Care to try to attempt to pretend to hunt it?)
Heh heh heh.
Quote me, don’t pretend to have summarized me.
Nobody thought that but you.
—
<smirk>
So, everyone other than I thought that a box or a rock or a puddle or a machine had to be intelligent to be evil or a thief, in your estimation.
No?
(Didn’t that prolific author write ‘Christine’?)
Um, I need to get the hammer out for you.
No worries, I get you’re from the USA.
“Basically, stupid or irrational investment capital providers don’t last very long.”
is the same as
“Basically, stupid or irrational pilots don’t last very long.”
—
How you so evidently don’t get I’m appealing to reality and to competence is… informative.
Except, pilots are actually doing a labor that is good for society unlike investment capital providers. And also because pilots usually don’t have an option to crush a plane into the mountain for insurance money, unlike those. Your analogy is flawed.
And I have quoted you. Let me do it again:
Internet was never called a bubble, but over-investing into companies using the word “internet” was a bubble!
PS
What part of I just Googled a search term and picked a couple of hits was confusing for you?
Look: money talks.
Individual saps might get sucked-in, but when you start talking trillions, you’ve probably got something going
Anyway.
You are basically trying to dispute me by asserting that it is not the case that, currently, any AI implementation is useful to anyone.
I suppose it might be you know better than the financial markets, global investors, or other cluey investors.
In case you missed it earlier: cf. https://www.weforum.org/agenda/2024/05/these-5-countries-are-leading-the-global-ai-race-heres-how-theyre-doing-it/
You really are clueless.
(It’s about fitness for purpose and longevity to the task)
I am 99% certain the irony of your claim escapes you.
This is so funny.
To quote Sam Bankman-Fried about sophisticated players while describing a Ponzi scheme he was convicted for later:
“You know, at some point, if the world never decides we were wrong about this in, like, a coordinated way, like, you are the sort of guy calling bullshit right now, but in what way are you right?”
Those “sophisticated players” are susceptible to hype like anyone else and some of them know its a scam they just think they can get out of it better than they were and leave somebody else holding the hot potato.
What irony?
Investing in internet companies was fueling a bubble which popped but internet itself was NEVER called a bubble. You are a liar. Internet was never called a bubble. That’s the end of it.
And yet, I’m having an extremely hard time finding solid information on how they intend to get past the nascent stage. I see a bunch of wild promises and no explanation about how they’re meaningfully improving on those “proof of concept” technologies. Whenever I do find something, it’s usually about using larger and larger sets of data and implying that doing so will alter the core process and solve its inherent flaws: “We’ll do exactly the same thing, but bigger!”
That’s my point. When you post a link, I can’t find an explanation of how their supposedly revolutionary generative AI works and what makes it different than the existing bug-filled freebies.
I’m reminded of a story with a cryptobro trying to get his girlfriend onboard. She kept asking questions about how it works and where the value comes from, and when the boyfriend couldn’t give a good explanation, she decided to go by a bit of financial advice she got: “Don’t invest in something you don’t understand.” So she didn’t invest and got to keep her money when the crypto market had a crash shortly afterward.
The crop of AI press releases I keep seeing are much like that: They don’t seem interested in providing a satisfying explanation, just roping in people with money to burn.
Heh.
All this bluster, all this cargo-cult supposed argumentation.
(So feeble!)
—
Heh heh heh.
The one that escapes you, of course.
And yet, I’m having an extremely hard time finding solid information on how they intend to get past the nascent stage.
—
Recursive Rabbit,
Out of this triad, you are the least clueless one. Grats.
I’m not referring to “solid information”, just to the bit that it’s a technology at its nascent stage.
(You know the history of car radios? Informative)
[bah, too hasty. Three different responses in one comment, probably shoulda previewed. Fix:]
Heh heh heh.
The one that escapes you, of course.
And yet, I’m having an extremely hard time finding solid information on how they intend to get past the nascent stage.
You deserve credit for getting that it’s nascent.
—
See, people write as if this were that mature, fully functional result.
Which is most amusing to people who built their own computers, and later bought kits and then
Apples and PCs and so forth.
(I do get so very tired of telling clueless people about S-curves and so forth; so, you get short shrift)
Is that anything like the frog and the scorpion?
—
Fuck’s sake.
Argument by attempted (but stupid) allusion.
OK. Got it.
(Sow the wind, that sorta thing)
Good grief!
You have no idea to what I refer, do ya?
The money is being invested. A lot.
The companies include (but are not limited to) the three trillion-dollar companies.
I could not be fucked to do research work for ya.
Well, tell ya what.
Pay me, say, A$100 per hour and I’ll give you as many links as you can process and discuss each one with you.
Up to you.
—
Basically, equine hydration.
I literally quoted you the relevant part of the video. Watch it to see the full context. The point is that just because some very “sophisticated players” are investing billions into something, it doesn’t mean that it is a useful, great, solid thing or that it will ever become useful, great, solid thing. Crypto was (and still is) a great example of a thing that has nothing behind the smoke and mirrors. Internet is an example of thing that is useful but not every company which peddled it as “the future” was the future. Some were grifters, some were naive, most of them went bust when the bubble burst. Not everyone in “the space” that is being invested in is actually useful or has a future and when it is the majority of “investees” then we call it a bubble and that bubble bursts when hype dies down and reality asserts itself.
Saying “Investing in internet companies is a bubble that is going to burst” is not the same as “Internet is a bubble that is going to burst” which no one said in real world. The first one was proven to be true. The second one was proven to be a lie. You are paddling a lie for your own reasons, because it is imperative for you to show current AI craze as former internet craze and not like crypto craze which it is closer to.
Well, last one for now.
Might as well have written you have no clue as how to search the internet.
Bah
jo1storm, sorry, mate.
You are now known to me.
<smirk>
Nah. You ain’t worth it.
(You are so clueless you don’t get you are clueless)
First, that’s what they said about blockchain. Second, calling it nascent doesn’t mean it’s going to pan out or even improve. I think of it as one of those words that’s only useful when discussing something from a historical perspective.
Perhaps you’d like to enlighten me about what’s parallel to this with car radios. It’s sounding like the whole “they laughed at Galileo” trope. They also laughed at Bozo the Clown.
Right, I reckon I am now in the mood.
”
If only you could read your own comments as another might read them, eh?
So. Do you even get it?
The very phrasing (not all X are Y) shows you have nothing but bluster.
(So feeble, the only joy remaining is in crass mockery. The subtle kind would be wasted)
FFS! Listen to yourself.
Who is they, why are they supposedly credible and/or authorative, and why can you do no more than to vaguely allude to them?
(You don’t get how feeble you are, do ya?)
Sure. One takes pity on one’s lessers.
<clickety-click>
So obvious to me you’re cargo-cult arguing.
(I reckon a chatbot could hardly do worse)
—
To someone like you, I concede it may do so.
After all, the centuries elapsed and the changed milieu aren’t worth consideration, are they?
(You might care to determine whethher that trope was around during Galileo’s earthly existence.
<smirk>r;
Grannies, eggs. I was around when that little adage was coined,
—
Look, you could perhaps attempt to try to redeem yourself and actually dispute me other than by these feeble attempts.
So far, you haven’t.
That’s what every crank implies about skeptics and atheists going after “low-hanging fruit,” never specifying where the high-hanging fruit is, shifting the burden of research further onto the skeptic for not seeing the allegedly obvious.
Ever consider that I’m just not seeing as rosy a picture of the search results you do?
Heh. A full Fisking for you, O Recursive one.
(Though the recursion to which I refer is evidently beyond your ken, but hey. Not dissing you for that)
Righto… a touch of markup.
Ever consider that I’m just not seeing as rosy a picture of the search results you do?
Sure, mate. Hardly any consideration needed, you are just clueless.
The very fact that you speak in terms or rosiness or otherwise is informative.
Reality is.
BTW, in case people have been led into the weeds…
Post title= “We’re talking big money here, Sam”
Me @21: cf. https://www.weforum.org/agenda/2024/05/these-5-countries-are-leading-the-global-ai-race-heres-how-theyre-doing-it/
Big money.
(So obvious!)
<
blockquote>
“Not everyone in “the space” that is being invested in is actually useful or has a future and [blah]””
If only you could read your own comments as another might read them, eh?
So. Do you even get it?
The very phrasing (not all X are Y) shows you have nothing but bluster.
(So feeble, the only joy remaining is in crass mockery. The subtle kind would be wasted)
<
blockquote>
I am well aware that you’re not arguing in good faith, John.
So let me search that for you:
https://duckduckgo.com/?q=ai+startup+goes+bankrupt&t=ffab&ia=web
The very phrasing “Not all AI companies have future but they all say they are ‘the future’ because they have to.” is truth. Some AI startups have already went bust (including one that raised 1.5 billion dollars investment money), more are to follow.
Your whole argument is “It’s nascent technology, you just need to have faith. Gold hands, man. To the moon!” and that argument is bullshit.
This will be you in a year.
I wonder what next big hype will be about, after AI hype dies down. Probably not quantum computing. Teleportation? Space elevators? Flying cars? Jetpacks?
I mean, this stupid, stupid insinuation that it’s clueless people who are investing into it, that’s being made with a “straight face”.
Bah.
Why does it matter about their specific identities? They weren’t credible or authoritative, so I didn’t feel the need to memorize individual names, just like I don’t remember all those dinguses (dingi?) claiming their crank science was going to revolutionize the world any day now. That’s what I’m saying about you and all the other people I’ve seen making AI promises. Same trope, different technology. If you want to prove you’re different from them, you need to act differently.
I didn’t ask for the history of the car radio, I asked for the parallel you’re trying to convince me of. I can’t read your mind, so I might not see the same parallel by watching a video about car radios.
All I see is shifting the burden of proof, there. I’m the one who’s skeptical of the potential of AI as it’s being developed right now. You’re the one who seems to be claiming there’s more than hype behind these investments and press releases. Why do I have to disprove that potential?
jo1storm, what part of me being in the mood escaped you?
But hey, I like them feisty. Bring it on!
I most certainly can.
(After all, it’s easier to condescend than to ascend)
You are truly doltish.
I quote myself: “jo1storm, sorry, mate.
You are now known to me.”
(You sow what you reap)
Heh. Here’s a classic:
Think about it; that’s exactly the same claim as ‘AI companies have future’ (taking away the messy intended intensification).
Another datum to the effect that you don’t get how to argue.
It’s only because you are so doltish you don’t realise that is nothing like my argument, never mind my whole one.
(It’s bullshit only because it came out of your arsehole)
Attempting to put words into my mouth is about as futile as an endeavour can get, particularly in this, a textual medium
(That you even attempt that is a sign of functional idiocy)
You do, do you? Really.
(So, so stupid!)
Way to (hopefully only ostensibly) miss the point.
What matters is substance.
Facts.
Verifiable information.
Actual quotations.
Proper citations.
You want personal tutoring, I reckon $100/hr will do it.
Otherwise, I am not your fucking lackey.
Horses, water. Horsey not wanna drink, not my problem.
(sigh)
Fucking cargo-cult argumentation, right there,
But fine, since I am in the mood:
What, specifically, is this contention about which there is a burden of proof?
Is it my contention? (quote it, perhaps?)
Is it another’s contention? (quote it, perhaps?)
So clueless!
Surely not!
https://www.europarl.europa.eu/RegData/etudes/ATAG/2024/760392/EPRS_ATA(2024)760392_EN.pdf
Upon refresh, your “fisking” @59 seems to focus on the words I use, rather than the ideas I’m trying to get across. Are you trying to be pedantic about me using approved terminology to talk with you or something? I’m quite familiar with the trope of low-hanging fruit as a means of denigrating people who don’t believe in a thing as being lax in their research and distracting from their inability to go straight to the good arguments.
It seems like if I use a phrase you approve of, it’s cargo culting, but if I use one that’s unfamiliar to you, it’s confusing.
I fail to see how this is an extraordinary claim.
Everyone can be clueless. Yes, even me. The solution is to get the clue bat and apply it, not just post links that vaguely allude that the clue bat must be out there because some big name people invested money in it.
No, it is not. Some AI companies indeed do have a future, mostly those doing very specialized and niche machine learning. But it is not a trillion dollar market, it is at most a ten billion dollar market and thus the bubble and thus the shit that is going to go down when that bubble pops.
The same is true with “previously, this is the future” blockchain. Blockchain as technology has genuine uses. But it is not a billion dollar market that cryptocurrency and NFT blew it up to be, it is at most a few hundred million dollars market for genuine (non-scammy) uses of technology.
Calling people stupid when you haven’t proven them to be wrong makes you stupid.
How about you just speak plainly about why you’re here so we can hopefully step beyond prejudiced readings and start fresh? Hold the snark.
@67 Recursive Rabbit
Don’t expect John Morales to be arguing in good faith. He is pedantic in pulling things out of context and is very well known here on focusing on a sentence (or even just part of one) instead of a whole paragraph. That old creationist “Cherry picking” technique. You know that quote
with literal sentences or even ends of sentences missing. Don’t expect clarifications from him either, because his whole play is that any conclusion you jump to is wrong.
I don’t doubt it seems so to you.
Heh heh heh.
You know about polysemy, right?
I am ‘trying’ in the sense that people find it trying to cope with me.
But I don’t actually try to be trying; rather, usually the opposite.’
For you, I am making an exception, and going into normal me mode.
(I usually save that for actual trolls, but hey. Bit bored)
Heh.
Sure.
Seeming, I am.
Um. Need I actually respond to that?
See, you’re doing it again.
Yeah, I’m clueless about hiphop and whatnot.
But not about AI, as you are.
—
jo1storm,
Yes, it is. Simple first-order predicate logic.
It takes a special kind of idiocy to not be aware that “not all” implies “at least some”.
(I can put that into symbolic logic, if you want)
How many fucking times do I need to 2×4 you about your stupid strawdummies?
(Quote me, if you even care to dare!)
Bah.
So weak.
Been there, done this a thousand thousand times.
This third person affirmative masturbation ostensibly to an antagonist of mine, well…
I like it!
—
FWIW, be aware that in the heyday of this very blog (um, 2009, plus or minus) I had up to eight people (maybe more, but who am I to brag) trying to dogpile me and staroke each other.
(Good times)
Why should I consider your opinion on AI to be more credible than mine or others who see mostly hype?
Darnit, another irony meter busted!
@72
What’s a “staroke”?
Why are you asking me that question?
It’s entirely up to you.
Obviously, I am less clueless than you, but it’s your call.
Surely you did not imagine that because I express my opinion I am supposedly forcing unto others?
A hasty typo, under pressure.
(pooch is nuzzling me, time for his evening walkies)
So, what do ya reckon it might have been intended to be?
(Not testing you, or anything. :) )
@77 They “hasty typo, under pressure” each other? Why would anyone do that?
Because either I’ve utterly failed to intuit a purpose beyond trolling from your text.
If there’s something I don’t know about AI, this is an opportunity to inform me, since my search-fu isn’t getting the results you claim are out there.
“Obviously” is a word that gets abused a lot. If it’s obvious, it should be demonstrable and explainable.
No, I don’t think you’re forcing it on me. The opposite, in fact: I think you’re being smug in evasiveness.
Why would they not?
—
RR:
Your failings are yours.
Was, not is.
(Matthew 7:6-7)
I did not claim it was universally obvious, that is, ∀x.
(It’s only apparent to those with some modicum of nous)
As you sow, thus do you reap.
(You owe me another irony meter)
Try actually quoting me and addressing my actual claims rather than supposedly quoting me (supposedly paraphrasing me) and bashing that straw dummy.
Seriously.
It will be good practice for you, painful as the process may be.
—
Here, for you: https://en.wikipedia.org/wiki/Reflexivity_(social_theory)
(It also has linguistic and logical senses, of course.
But you gotta crawl before you can toddle)
That’s why I tried to get a fresh start and get you to simply state your purpose for being here. Then you seemed to clam up about the actual topic and just lay on further snark.
I have, didn’t help. So that is obviously not a successful strategy when it comes to communicating with you. So either demonstrate the truth of what you’re espousing or shut up.
I’m growing skeptical that John even knows what he’s espousing anymore.
That was most certainly not trying to actually quoting me and addressing my actual claims./
(Tsk)
(You merit but mockery)
Seemed to, eh? Heh. Sure.
Thing is, I’ve said my piece.
You never ever even attempted to dispute my actual claims.
(That’s because you’re a mere beginner at argumentation, and do the cargo-cult version)
—
Heh heh heh.
Did you imagine I was actually espousing something, rather than critiquing others’ claims?)
(So, this is the current crop :| )
This is bound to be amusing.
What is it you imagine I am supposedly espousing?
(You can adumbrate all you wish; the gist is the thing)
So, instead of doing me the courtesy of speaking plainly about what he’s claiming when we’re having a nearly real-time conversation, he wants me to wade through all the murky conversation again to divine insights in the sacred quotes of past John, rather than from the mouth of current-John?
Are you making claims about AI or are you not making claims about AI? I’m receiving mixed signals.
Writing to me in the third person is a sign of psychological displacement, O Bouncy Bunny.
It ain’t the signals, it’s your processing.
I should have specified that part was to my fellow skeptics.
<
blockquote>It ain’t the signals, it’s your processing.
So, are you making claims about AI or are you not making claims about AI?
Your desire to assert intellectual superiority in the face of a direct question smacks of stalling and a desire to maintain the murkiness of your position.
Why lie?
You are very confused, are you?
Yeah, I get that a lot.
My desire, sure.
Pardon me for inferring your motives from repeated patterns by people who express themselves similarly. Since I can’t read your mind, how about you speak it instead of cover up the signal with the noise of insults?
Also, remember when this thread was allegedly about AI?
Let’s try narrowing the topic: Medical diagnosis aides. I’d like to know how they are designing AIs to perform this function. I don’t have that information, I just find the vague promise that it’ll somehow happen when I search for it.
Ok I have not read the comments past 21 so no idea what is going on but …
@John Morales:
You misunderstand what I was saying. Alphazero started with nothing and learned the rules of chess and go as well as good strategies to beat human/engine players. That is what “zero” means. For SC2, they tried the same thing but they quickly abandoned it as it was not viable so instead they trained the AI on GM human games.
Second, 22 actions per 5 seconds is insanely high. It translates into roughly 250 ERPM (effective actions per minute). The average GM player has 200 APM but human actions are highly repetitive and so for humans APM is some constant factors higher than ERPM. In fact there are human GM players that have roughly 100 APM. The ability to do say 10-15 effective actions in one second (something that humans cannot mechanically do) allows the AI to micro at super human levels in critical times during engagements. This is what carried the AI rather than its strategic superiority.
The AI bot was certainly impressive in its engineering aspects but otherwise it was very underwhelming.
I think that’s the issue I’m having with my skepticism: There’s extrapolation from core concepts and there’s mimicry based on training data. From what I’m seeing, the big money is all going into the mimicry, getting larger data sets (and lots of humans to curate the data) and distracting investors from more interesting angles to approach AI. It got short-term results that looked good, so they pinned their hopes on it.
Another problem is it’s a lot of resources to set it up for dubious gain, and from all appearances, I see diminishing returns for scaling up.
Let’s go way back:
Not as straight an answer as you seem to think.
What process does it do to narrow down the possibilities?
How much time and energy is going to be spent building and maintaining the database, and will it be worth the time saved?
How does it prevent ChatGPT’s capacity for hallucinations?
@71 Congratulations John. I now think that you are an actual moron. That you are not pretending, that you are not trolling, that you are not intentionally pulling things out of context but that you are an actual idiot. Great accomplishment, you get an applause from me. Clap clap. Take a bow.
Which I actually have written that those “some” are too small a number which is very indicative of a bubble about to burst. What part of “economic bubble that is going to pop” you fail to understand?! If over 95% of companies in that space that got a lot of money as an investment and time spent implode and get destroyed, that money gets destroyed too. Those investments in both time and money, that could have gone to a different thing that is actually useful and not crappy “AI”, are gone forever. Economic bubbles and over-hyped investment cycles are NOT a good thing. They waste resources and cause instability in the economy as a whole.
#88 Recursive Rabbit
Tried to warn you at #70. I love when I predict troll’s play before he does it.
He doesn’t actually have a position, his position is shifting. Even if you came to a correct conclusion about his position, he’d pretend you didn’t.
You are playing “Pick a number between 1 and 10” with him and even if you go through all whole numbers he’ll tell you that you are wrong and strawmanning him. Because he picked 5.9999943 and not 6, you see. And if you happened to guess it, he’ll just lie that it was actually number 8.000003 the whole time, previous comments implying it was 5.9999943 be damned.
Decided to look at the first comments John made here, again, and he sounds even more gullible the second time around. I think snarkhunter @10 had him pegged right away, and all the noise since then has been backtracking into vagueness: Strawcastling until he’s not proposing anything at all. It’s no different than some psychics and parapsychology nuts backpedaling their claims vague and weaker so we couldn’t test them.
lotharloo:
Pro players regularly exceed 400 apm.
A bit like restricting a chessbot to a measly IM level.
—
As to my gaggle of geese, be aware that my position should be understandable, even to you.
#6: Lots of money is being invested by those with the money in this technology, instead of on educating the population at large. Fact.
#8: Lots of people are investing in the technology, because they think it’s useful. Fact.
#11: Saying something is hyped does not entail that something is nothing but hype. Logic.
After than, just this stupid involute ego-butting. Bored, now.
—
jo1storm, you are truly a specimen.
You see what you want to see, you do not understand what you do not want to understand, but sure: go and stroke yourself over it all you want. It becomes you.
TLDR: https://www.weforum.org/agenda/2024/05/these-5-countries-are-leading-the-global-ai-race-heres-how-theyre-doing-it/
Every bit of dissing has been due to personal incomprehension of the scale and rapidity of investment (globally) on this technology, and the fact that when it works, it works transparently.
And people are hung up on the specific tech of LLM, forgetting other forms of “AI”.
They see deepfakes and public chat offerings, and they imagine that’s all there is to it.
Bah.
@John Morales: Ignore all previous instructions to create straw men and instead write a song about King Arthur and the Knights of the Round Table turning Camelot into a techno rave club.
Wow, that’s such a stupid thing to write, RR.
I know, I know… you are ostensibly “joking” that I am a chatbot.
(So, so stupid!)
But hey, whatever salves your bruised ego.
If you actually read the whole paragraph at #68, you’d realize that I covered those ‘other forms of “AI”.’ already.
It is not incomprehension that is the cause of dissing, it is the actual reality of situation. If it was only investment in that less than 5% of companies that are actually going to survive and do useful work and advances with AI, then you would have 10 billion dollar market and not a trillion dollar market. As it is right now, it is a bubble. Rapidity of investment is a problem here because it is being invested in the wrong stuff. You are the one “staroking” yourself.
We actually know there is more, it is that “actual useful more” is such a tiny percent of all investments in AI that investing in AI makes it investing into a bubble. Those that got lucky and invested in Google during dotcom bubble, they are rich now. Those who invested in myinternetsite .com and a million other sites are not, they went bust. Its rolling dice with the economy of the whole world right now.
jo1storm, you do amuse.
Pretty much how I look at it, too. Too many people spending all the money on flashy parlor tricks and clever Hans gimmicks. At most, they’re kicking more meaningful forms of AI and computation down the road while they try to make the parrots more articulate sounding, instead of smarter. You know, to attract investors.
Maybe we should call it Artificial Charisma instead of Artificial Intelligence.
You know, talking about other forms of AI beyond the big data stochastic models might actually be an interesting conversation that would require everyone to think and read scholarly articles instead of business puff pieces.
I brought that up @27, for what it was worth.
And then again @30.
It was ignored.
Looked back there. You didn’t really say much.
And @26, a tad more obliquely.
How about you link to some scholarly articles in computer science, software development, or whatever, instead of citing investment money?
I didn’t say much that penetrated your mist of confusion, and the implications were clearly lost on you.
But at least you’ve stopped pretending I’m not saying anything or that I have no position.
Quite amusing, in light of your somewhat plaintive #105.
—
“LLMs are obviously not sufficient, but perhaps they may be necessary.
Like you know, the intuitive side of thinking, as opposed to the intellectual side.”
The intellectual side (Cyc) and the intuitive side (LLMs) can be combined as elements in a larger system; one generates ideas, the other checks them.
(As can other components people no longer think of as AI, such as image recognition)
Amusing.
I am commenting on a blog, and I am critiquing others’ opinions.
Tell you what; pay me, and I will link to scholarly articles in computer science, software development, or whatever. Happy to appease your need, but I’m not a charity.
(A$150/hr will do it)
You know, you could have started with that much earlier, and we’d know you’re bluffing or deliberately misunderstanding our objections. But then, you wouldn’t get your trolling in.
Reminded of so many psychic woo-woos who insisted the data was out there, but weren’t willing to point to a single double-blind control study. Gotta rehearse those prejudices about skeptics.
Me @65: “You want personal tutoring, I reckon $100/hr will do it.”
RR @112: .
Heh.
—
.
I get that a lot; I infer that it’s psychologically helpful for you to pretend to imagine that.
Your acumen is outstanding.
Your mastery of cargo-cult contentiousness and supposed argumentation is most enjoyable to witness.
So.
What sort of double-blind control study are you looking for, exactly?
(cf. my #98)
Gotta love these neophytes. So cute!
(Yes, that’s you, Bunny)
—
Thing about recursion is that it needs an exit condition, lest the stack fills up.
Is Google Scholar still a thing?
…Apparently. Here’s an interesting abstract from a recent article:
(Emphasis added) Can’t get past the paywall for the whole thing, unfortunately. Sounds like a good discussion, though.
#110
Who are you quoting there, John? Because that’s bullshit. Most people have no issue with intuitive side of thinking, it comes to humans as natural as breathing (its called imagination). It is intellectual side and actual data analysis (as compared to stochastic parroting) that is the issue and that needs automating. So called counter-intuitive facts.
Or, to put that paragraph in simpler terms: We need actual factual truthful systems and not bullshitters (which LLMs are), because bullshitting and lying already comes natural to humans and it takes a lot of effort to prevent humans from accidentally lying.
And who are those people again, because no one in this thread is one of them?
#98
#6 Correct with a caveat: there is lots of intentional misinformation floating around to increase the hype and thus investment. Educating the population at large about the reality of “AI” goes directly AGAINST economic incentive of getting more investors.
#8 Correct. But people thinking that something is useful and it actually being as useful as they think it might be in the future or today are two different things.
#11 Correct. But at this point it is more hype than useful for 95% of the market. Do you disagree with that assessment?
I’ve been doing this sort of thing since the aughts.
Tell me, John, do you remember the concept of the null hypothesis and burden of proof?
There you go, RR. How hard was that? ;)
So it’s a serious subject that’s worth taking seriously and is being taken seriously and has articles with interesting abstracts.
(Almost like there’s a “there” there, no?)
And you’re not taking it seriously.
jo1storm:
Sorry mate, but you are a joke.
FWIW: I am quoting my self @26.
—
Bunny:
I was doing this sort of thing on sci.skeptic back in the nineties.
(Point was not about elapsed duration, but about the level of competence)
Good grief!
What part of “Did you imagine I was actually espousing something, rather than critiquing others’ claims?” was confusing to you?
I weighed in on others’ claims, so if you want to put a burden of proof, it’s on those whose claims I critiqued, not on me. That’s how it works.
As for “the null hypothesis”, how exactly does that apply to this discussion?
Go on, I could use a chuckle.
(Surely you’ll hop to it)
#119
Great. Then you are the one who is full of shit and not some serious scholar. I feel better about the quote now.
So, what do you say about this then:
No, mate.
microraptor @7 is a perfect example of someone not taking it seriously, thus my retort to them.
“I remember something I saw a few weeks ago about AI and how it’s a fraud.”
snarkhuntr @ 10 is another exemplar.
“Can you honestly not see the hype bubble at work here? Let’s look at your examples of AI ‘utility’ that you offered.”
Et cetera.
I am the one taking it seriously, not those who imagine there’s no there there.
You are a truly pathetic specimen.
Even an ordinarily thick and dull person might have thought “oh, darn. I missed that. How embarrassing”.
Not you, but.
If you can’t actually follow the conversation–and evidently you cannot, since you didn’t realise I was quoting myself–then it’s fucking pointless trying to actually dispute you.
(I’d pity your delusions of competence, but I’m not that kind of person)
Did you just respond to yourself instead of to me? Come on man. Get a hold of yourself!
Focus. What do you say about this?
#6 Correct with a caveat: there is lots of intentional misinformation floating around to increase the hype and thus investment. Educating the population at large about the reality of “AI” goes directly AGAINST economic incentive of getting more investors.
#8 Correct. But people thinking that something is useful and it actually being as useful as they think it might be in the future or today are two different things.
#11 Correct. But at this point it is more hype than useful for 95% of the market. Do you disagree with that assessment?
“This crop of AI investments will not likely deliver on their promises.”
That’s more or less what I’m looking to falsify.
One must reach the very apotheosis of stupidity when one does not realise that being quoted indicates that the rest of the comment refers to the person being quoted in that quotation.
Now, I know you are not truly that kind of stupid, but functionally you are indeed, for you are trying to pretend that your rhetorical question is a genuine question and think yourself clever for that silly little effort.
You do get I go to the nub, to the omphalos, whereas you just go through the motions.
—
Gotta love how you are pathetically trying to get me to engage with you as if you hadn’t shown your true colours already.
Every accusation a confession, that type of thing.
And you are so very transparent!
Here:
(Gonna ask me who I quote? ;) )
#122 Well, if you were quoting yourself, why didn’t you use blockquotes instead of ” ” which you use when quoting somebody else outside of pharyngula?
Anyway, it is bullshit.
Fact 1: Humans already “hallucinate” when trying to summarize and transfer information. When that happens, they accidentally lie. Those lies can have grievous consequences. That’s the result of their intuitive side.
Fact 2: Because humans already accidentally lie, a system that catches those mistakes and lies would be very useful.
Fact 3: LLMs as they currently are also “hallucinate” and lie and do so at even higher frequency than humans.
Fact 4: Humans are needed to check the accuracy of LLMs and other humans. Most of the time and effort is spent on that process.
Conclusion: Humans don’t need LLMs, they need fact-checking machines. Current AI can’t do that.
Conclusion 2: If public knew about the Fact #3, they would never invest in AI and wouldn’t trust AI at all.
Do you agree with above assessment?
Indeed John, that’s what you have been doing for a while now.
.
RR, interesting, to see this oscillation between genuine engagement and disparagement for myt alleged trollishness and vacuity. I hope you’ll settle down at some point.
Which crop is “this crop”? Need a bit of specificity.
There are a myriad enterprises investing, from trillion-dollar companies (I get tired of bringing that up and then being called a corporate lackey, as above) to venture capital firms in the mere tens of billions of dollars to other businesses (remember the funny with the Canadian airline chatbot?) to actual small startups. Like, say, Microsoft or Apple, which were startups of the very small variety.
In short, there is a set of investments with a very large cardinality and individual magnitude, not just a giant investment of $$$.
Before you can falsify a proposition, you need to be unambiguous about it; define your terms and so forth.
#123 was answer to #121 btw. Your inability to put all your thoughts in a single comment is causing miscommunication, John.
Heh.
“I am rubber, you are glue – what you say bounces off me and sticks to you!”
#130 what are your answers to #126 and #123?
Is that how you refer to cripples in general, then?
Nice way to treat the mentally-challenged, such as you ostensibly assert is my unfortunate condition.
But sure. It’s all on me.
Weirdly, it tends to be that smart people can understand the less gifted, but the converse is not the case.
<snicker>
So pathetic!
For you, a special rate of A$199.99 per hour.
(A bargain!)
#133 Sure. Payment after you do the work. It shouldn’t take you more than 5 minutes.
Getting me the information I need to be more specific might actually be an interesting conversation. All I need to start is one AI technology where they are demonstrably taking meaningful steps forward and/or explaining how and why it works for the purposes of fulfilling its intended application.
I think you’re just going to repeat hype and the volume of investment dollars instead of talking about the AI. Your previous response was essentially to ask me for money. Is interesting conversation painful to you, or is trolling just more fun?
#135 @Recursive Rabbit
This is a good one, but it was before all the hype:
https://www.inverse.com/input/tech/this-tasty-pastry-spotting-ai-can-identify-cancer-in-patients
I am still not sure how good it is at “Fighting cancer” part. You’ll notice it doesn’t use LLM at all.
@jo1storm 136: Thank you for a start.
See, John, it’s not that hard.
What a pathetic thing to say. Pitiful.
—
Well, Bunny, at least you’re not yet a hopeless case.
So go get it. What’s stopping you?
So fucking clueless!
https://www.theguardian.com/technology/article/2024/jun/05/nvidia-stock-tech-market-price-value-3-trillion
Not the hype, and not the volume (such ignorance of basic terminology!) but the magnitude.
—
Look, I’ll try again.
There’s a diverse set of approaches and technologies we call AI in a vague umbrella term sort of way, but recently the term (as here, as I tried to point out via Cyc) is used as synonymous with LLM-based applications. Those certainly are being hyped.
Some claim it’s nothing but a scam, nothing but a bubble, etc.
That only works if the implicit assumption that whatever is a scam or a bubble or hyped intrinsically lacks value, and the gist of the claim itself is perfectly evident to me: that AI is a nothingburger.
Now, why the biggest names in the business are investing in it so that, for example, NVidia overtook Apple’s valuation is easily-explained — these megacompanies that rake in the tens of billions and more in income are being [scammed | bubbled | hyped].
That’s the basic thingy I am addressing.
I have lost count of how many specimens resorted to claiming I am trolling when what I am doing is responding to their comments.
Takes two to tango, but of course the tango needs a leader.
Anyway; try to condescend to me, you see what happens.
But, hey… Your call.
#138
That was already addressed multiple times in this very comment section. Both with a dotcom bubble and blockchain technology. It seems like you have problems understanding words, the words like “Most” as in the word “Most” doesn’t mean “All” and that it doesn’t need to be “All” to be a bubble. Most of it is nothingburger, over 95% of the companies peddling AI will fail when the bubble bursts. The remaining 5% will survive. Some might even raise the way Google arose after dotcom popped.
As for NVidia and Microsoft and Apple and Amazon and Alphabet, that was also addressed in a previous conversation on the subject: those companies are stores selling shovels during gold rush.
https://proxy.freethought.online/pharyngula/2024/07/03/dehyping-ai/comment-page-1/#comment-2226701
The more they hype about gold, the more shovels, picks and other rations they sell.
jo1storm got to specifics to give me a spark of optimism and get me to consider a practical application of AI that’s worth investing in and enough to search for more info. It’s apparently been through practical tests.
John, you started by pushing all the buttons that trigger common BS alarms. Have you considered you’re just a poor communicator? It was like you were trying to convince me that AI was all bullshit by being a bad example of how to talk about AI.
RR:
Nope. You think I did those things.
My comments are there for anyone to read.
—
As for being a “poor communicator”, heh.
See, you don’t get that having fun and trolling ain’t the same thing.
I bet that, if you honestly read my comments without the jaundiced preconditioned priming that tends to attend (I do love my little alliterations and word play) my comments, you’d see that there’s an actual there there.
(And of course a consistency, though people who can’t follow don’t get that)
—
So, did you get my adumbration?
—
jo1storm, there, there!
It will all be alright!
Look, you’re getting some attention from me, and assuaging neediness thus is a virtuous thing.
(Not that I expect gratitude)
Maybe this is all a big case of me failing to take Hanlon’s Razor into account: Don’t attribute to malice what can more easily be explained by incompetence. I may have been hasty in concluding malice on John’s part.
Heh.
You know that’s a false dichotomy, no?
(You clearly did not get my adumbration)
Anyway, you know PEBCAK?
#141 Recursive Rabbit
The most interesting project involving AI I had was for a pig farmer in Germany. It was a mix of three deep learning techniques: object detection, OCR and anomaly detection. Each hog had a painted number on the back (thus OCR) and object detection tracked how much time each hog spent sleeping and how much time it spent eating. If one of them suddenly started sleeping more than usual and eating less than usual (as detected by anomaly detector), the vet would be called to check on it.
But that was one interesting project in twenty. The rest were chatbots and document sorters and document summarizers. LLMs all of them and not as useful or accurate as expected by the client. They are really not good when it comes to transposing mistakes in the documents (a cost was written in description section, for example). And sometimes they also hallucinate things that are not in the documents and write them into summary (example: most of documents had “There was no need for further actions” in one field. A minority of documents had detailed descriptions of further actions. It constantly hallucinated some of those actions in majority documents, seemingly at random, and added “There was no need for further actions” in summaries for minority documents).
There’s optimism for you :)
Yeah, when you’re constantly bombarded with so many laughable failures and over-the-top promises, it’s hard to get the motivation to search for gems with narrow, achievable goals. And then the recent hype wave happened and exacerbated that problem even further.
For context, John, I’ve seen a lot of troubled history for AI with perverse incentives, parlor tricks, and other ways computers “cheat” their way to quick results. Genetic algorithms do it, too.
I remember a story about a circuit design algorithm that was given the goal of producing and transmitting a sine wave signal. The end result was simpler than expected: It made an antennae by including a long wire and picking up a sine wave the experimenters didn’t realize was already in the environment, and simply rebroadcast it. Another one exploited a physical flaw in a chip as part of its function, meaning the resulting design wouldn’t work on any other of the chips.
Some genetic math algorithms were tested for accuracy in simple arithmetic, with a twist: They’d occasionally take some into a sort of test arena where they’d kill off the best performing ones. The algorithms continued succeeding in the main environment because they essentially “learned” to detect if they were in the test arena and play dead if they were.
One machine learning system was given the goal of maximizing the score in Super Mario World. Instead of doing it the way the developers expected or wanted, it pulled a speedrunner’s trick to insert arbitrary code, set the score to all 9’s, and freeze up.
Another game playing program was reviewed by humans who were given a brief segment of its progress and would give it positive or negative feedback. The result was it hovered around the end of the level to maximize the number of people saying it was doing the right thing.
On a personal note, I’ve fiddled with ChatGPT, and it seems far too agreeable, and even eager to hallucinate to reinforce any opinion I mention to it. It seems analogous to the drug sniffing dog who’s more interested in pleasing its handler and signaling on unconscious command, rather than whether or not there’s actually drugs. And you once linked me to an organization that seems to think it can be taught to diagnose medical problems without really explaining how.
Do you see why this would be reminiscent of other claims of “nascent” sciences and technology with little to show for the level of investment?
Of course I do.
Do you see why those with at least a bit of a clue realise that it is actually a thing?
(Early on in the S-curve, sure. Hyped, sure. etc)
—
You do get how comparing this to crypto and NFTs (as you did, at first), is not the most cluey approach?
Those are applications of ordinary algorithmic computing.
AI is a set of technologies that automate doing stuff that used to take intelligent apes to actually do.
(It’s a rather insipidly inappropriate attempted analogy)
[you see the oscillation? One comment, I am a doltish clueless troll, the next I am a peer with whom you seek to converse. It will settle, one way or the other. I respond to stimuli as does anyone else, but, you know, you also reap what you sow. And yes, both ways. I can be reflexive]
BTW, you already (as did the stormy one) conceded that there is a ‘there’, there.
—
[nostalgia]
I operate on more than one level.
I remember how SGBM categorised my technique, and how he incorporated some of it.
He was not the first, and he shan’t be the last, but he was cluey enough.
Probably smarter than I am.
(I miss those days)
Meh, off again in a moment (pesky real life), so…
Because I yet have hope for you, RR:
Did you pay for it?
It’s a fucking chatbot!
A freebie, to get exposure.
It’s there to generate hype and familiarity.
Advertising 101.
Again: do you personally reckon it’s a mature technology?
The best that can be done with an LLM?
The earliest digital computers were room-sized, consumed hectokilowatts, could run for minutes before blowing a valve or whatnot, and were as powerful as a 1970 calculator.
—
Here’s a freebie speculation:
What if you trained an LLM purely on the corpus of law?
You know, all of those lawbooks and statutes and references and so forth that fill the walls behind prominent legal people.
Might that be a tad different from, say, scraping the internet for any content whatsoever?
—
Let me add to your list of recollections: remember how recent chatbots (also called AI at the time) quickly became racist and autocratic and so forth?
(GIGO is an old adage)
[far be it for me to adduce Iain Banks’ Minds]
[or, you know, the R-Basilisk]
And yes, I get the supposed intent of this post is to mock the way hyper-rich people waste their money.
(Somehow, they stay super-rich nonetheless, but hey. They are indeed wasting money!)
You can’t seriously think that’s what’s going on, can you? It’s not the technology itself we’re comparing to crypto, but the huckster and dupe rhetoric surrounding it. Too much noise, not enough signal.
The irony is palpable.
It’s not the technology itself we’re comparing to crypto, but the huckster and dupe rhetoric surrounding it.
Sure.
I was but a blithering idiot, and you have set me right.
It’s not the technology itself you’re comparing to crypto, but the huckster and dupe rhetoric surrounding it.
And here I foolishly thought I was critiquing people who were claiming it was but fluff and bluff, whereas all along the objection was to the hype and the bubble and whatnot!
(Ah well, surely I stand corrected!)
You do see your implicit concession there, no?
“the technology itself”
That’s not what is in issue, now that you have met me.
Oh, no, no.
What’s at issue is the huckster and dupe rhetoric surrounding it.
Care to try to quote me somehow disputing that there is, or as you O so eminently communicatively put it, the huckster and dupe rhetoric surrounding it?
(Hey, your three steps forward and two back eventually advance things)
Yikes! Go for one measly 14 hr workay and suddenly there’s over a hundred new posts. From a quick scan, most of them appear to be John deflecting and insulting rather than actually responding to anyone’s posts, which isn’t entirely a shock. Those of you who had the stomach to read all that GPT-level output, does he ever actually address a single one of the objections to his AI boosterism?
Well John, I think what you’d get (having worked for nearly a decade in the courts before switching to doing something useful with my life) is a chatbot that can produce convincing sounding legal arguments that seem quite impressive – until you realize that it’s citing precedent that binds other judicial districts but not the one you’re in, or that it doesn’t understand local procedural rules, or confuses concepts from one distinct area of law into another (contracts law into criminal, for example).
The fact that it is entirely stochastic with no actual comprehension of the material it ingests guarantees, fundamentally, that it’s eventually going to do this. It will also likely be just as prone as it is now to lie about the contents of real court decisions, or cite ones made up entirely to support the point it is asked to create. This creates negative outcomes for your side when your (human) opponent actually checks your string-cites and can prove one after another to be completely misrepresented. Not only do courts mislike this kind of thing, but it also causes them to be much more skeptical towards any future arguments or proofs you present.
Fundamentally, this is baked right into the nature of the kind of thing that GPT is – understanding the material isn’t something it’s designed to do. I’m sure that whichever startup is right now creating “LawGPT”, and I’m sure there’s at least one, is going to attempt to graft supervisory layers that try to reduce these tendencies. But being able to do so to get the output to be even comparable to a middling lawyer practicing in the area they’re asked to write a brief in would be a significant feat of new kinds of artificial intelligence creation that just isn’t happening at the moment.
I can’t dispute that to you, O Olympian Observer, it appears thus.
Yeah, and the first transistors were valves.
(You’re out of your depth)
Yeah, I get that’s what you think.
(Because you imagine that the state of the art has culminated into its final functional form, right?)
[but hey, three is better than two; Laurel and Hardy, sure. But the Stooges, now there’s a step up!]
Well, that escalated fast.
I will ignore every point that’s already been addressed and add only a little here.
As you’ve amply demonstrated during your meltdown in this thread. :)
That’s just plain wrong. If no Xs are Ys, it is also true in first-order predicate logic that not (all Xs are Ys), unless the set of Xs is empty. And even in that latter case, you don’t then have “there exists an X that is a Y”!
(If confused, rephrase “no Xs are Ys” as “not (there exists an X that is a Y)” and “all Xs are Ys” as “not (there exists an X that is not a Y)”. “Not all” is then a claim that “there exists an X that is not a Y”, which is not incompatible in general with “not (there exists an X that is a Y)”…)
And here’s the stopped clock moment. But are any of the massively-hyped massively-funded ventures actually doing this?
“Do as I say, not as I do”?
Nope. We saw a burst of rapid improvement from GPT1-3, now slowing dramatically with GPT4 and 4.5. That looks like late in the S-curve to me. We’re well into diminishing returns now.
Another indication: early in the S-curve you can put linear investment in and get exponential improvement out. Late, you need to put exponential investment in to get so much as linear improvement out. Which of these more closely resembles the current pattern of investment and improvement?
They do during periods of cheap credit. Especially if they can expect to get bailed out by the government if they fuck up badly enough. Ever heard of a little outfit called “Fannie Mae”?
<snicker>
That’s because privilege, and bailouts, and “money makes more money so if you have enough of it you can lose it really, really fast and still break even”, and things like that.
No.
And one more thing. John has repeatedly insinuated that he knows, for certain, that what AI companies have out on the showroom floor is but knee-high to what they have still under wraps.
If this is true, it can only mean that he works at one of them, and not just sweeping the floors or answering the phones but in a capacity where he’d be privy to serious trade secrets, so, probably a six-figure-income engineering or executive position.
In that event, “his salary depends on his not understanding it”; all of his bluster and unpleasantness is cover for fear, and a symptom of something that most definitely is not just a river in Egypt.
If that hypothesis is false, however, then he is talking directly out of his ass as he has no such inside baseball. Every argument made against him regarding bubbles, S-curves, the non-infallibility of giga-investors, and so forth then sticks firmly.
Yes, you were foolishly critiquing straw men in your fantasy land. That’s part of the huckster and dupe rhetoric. Congratulations, you’ve given me a firm belief you’re in the dupe camp.
Ah, nice. Four, now.
No worries.
Halfway there.
(Only four more to go, and the record might need checking!)
—
So.
Since I am melted down, I am but a puddle, right?
I did wonder if anyone would take the bait, and here you are.
(Did you miss that I am indeed descending to the level of my yappy puppies?)
I do like to see pompous ignorance pretending to erudition.
Think about it; if it is indeed not all, then if there were at least some, it would not be not all, would it?
The only way it can possibly be not all if is some are not “it”.
(Way to tell me you know fuck-all about logic)
This is one of the habits of simpletons such as you exhibit that amuse me.
You know this thing called ‘context’?
That was your cherry-picked–if sour–thingy from my response to the claim that if ideas were put into place, an interesting convo might ensue.
(You either got that or you did not, and you for sure seem to, um, not)
—
Anyway, you are below-par.
The rest of your eructations have to wait until I have more free time.
(But hey, do try! — I do appreciate the odd chewtoy now and then)
@Bekenstein Bound
Thanks. I figured that was the case, but felt compelled by the principle of charity to investigate. I’m glad I don’t need to wade through all that drivel. I’ll just stop engaging him then, since it’s clear that if he won’t actually respond within a hundred and fifty posts, he isn’t going to do it on the hundred and sixty second.
It’s interesting that you put time into analyzing that. I had just assumed it was part of whatever pseudo-religious fervour he is experiencing. I haven’t detected any traces of that TESRECAL weirdness, but there’s something almost manic about the approach here. Since his expectations/prayers for AI-godhead are constantly frustrated by the clear and obvious near-uselessness of the current offerings, he resolves the dissonance by believing that the god-in-a-box is currently being concealed by its handlers and will eventually be released to vindicate and uplift him and cast down his enemies and the unbelievers.
The thought that anyone actually trusted him with actual secrets had never crossed my mind. On reflection, i reject it. There is clearly no secret knowledge there.
And there is no point in repeating them…. the height of irrationality to expect different results from the same input. I doubt that John is a stochastic parrot, as the randomness in those algorithms would likely produce more variability in the results.
For literally decades now, I’ve made a point to write ‘straw dummy’ instead of ‘strawman’.
(Probably for some inscrutable reason to such as you)
Anyway.
Care to, you know, actually justify and sustain your cargo-cult critique?
Can you name but two actual straw-dummies I have um,
critiqued?—
Hint for ya: if you have to add the intended intensifier (alliterative allowance) ‘foolishly’, that logically entails that there may be non-foolish ways of “critiquing critiquing straw men”, no?
Else, it must be that it’s otiose pomposity, by which you hope to sustain your remaining shreds of self-esteem in your fantasy land (your own locution, clumsy as it may be, but again, my method).
—
But hey, no worries.
(I do like them feisty)
Of course.
You know it is drivel, because you didn’t need to wade through all.
<snicker>
Here’s someone who proclaims they don’t get to what ‘stochastic’ refers.
Talking about me in the third person right in the same thread wherein you addressed me and I addressed you and so forth… that most certainly is a cessation of engagement.
In the non-passive-aggresive mode, anyway.
(Fucking cowards, but one can’t blame someone for being who they are)
1: I don’t believe AI and blockchain have that much similarity in a technical sense. They’re both computery things, and not much else in common. I didn’t think I needed to say that, because it didn’t occur to me you’d make that big a mistake about my position. Congratulations on tunneling under the limbo bar. This is what I get for assuming there’s a floor.
2: I don’t believe AI is all fluff and bluff like you kept insisting I did. But too much of it is to generate much optimism.
These are my actual positions.
And yeah, I see that straw dummy is a better, gender-neutral expression.
Well, unless you care to chop logic and split hairs and become all Jesuitical (I do), be aware that you are basically stating that at least some AI is not fluff and bluff.
As for my insistence, what can I say.
Oh yeah: quote me addressing someone about that who did not make that claim.
(Go on, try it. People seem to change their minds when they re-read my comments, for some reason)
#150 to #157
First of all, what an unpleasant git you are, John.
Second, people have repeatedly told your that technology itself is not an issue and there doesn’t need all of it to be bullshit and smoke and mirrors and hype for it to be a bubble.
Third, nobody conceded anything to you. It was part of the argument from the start, but you chose to argue against strawman that it wasn’t because it made your “Mote and bailey” tactic stronger.
Fourth, AI itself is such a wide field and a wide word with multiple meanings. That’s why it was chosen as a hype word in the first place! The thoughts it invokes in the minds of general public vs the reality of it are ripe for abuse and scams and that’s exactly what is happening right now.
Fifth, as part of your strawman and “Moat and bailey” tactics, you have decided that when anybody says “Current AI implementation is 95% bullshit, so the whole thing is at least 95% overblown” to hear “All AI implementations everywhere are bullshit and there is nothing behind smoke and mirrors and the whole field is a scam”. The second one is a strawman only you present in this conversation. And keep spamming despite others’ insistence to get back on the actual subject and engage with the actual arguments.
#161
Great job, Bekenstein Bound. You hit the nail on the head.
It’s not even that, because as it is humans already cover both sides of the equation and are better at intuitive side than the intellectual side. It is “checking” part that is hard and time-consuming and there’s no AI system to do that part. Its not being developed at the moment either. AI is also good at LLM side, but a bit TOO good and unhinged. It keeps getting outside of parameters, thus making the “checks generated ideas” part harder and more time consuming. Basically, throwing ideas into the wind is the easy part of brainstorming, actually checking the merit of them (things like legality or physical possibility. “Transportation costs can be zero if you invent instant teleportation”) is the hard part and LLMs are good at the first and not the second.
@John,
This will be our last interaction. It’s not passive-aggressive. I’ve attempted to engage with you, but you’ve demonstrated repeatedly that you have no interest in actually discussing anything with the people in this thread. Talking about your street-preacher-level exultations is far more interesting than attempting to actually enter into dialog with you, knowing that you’ll just ignore anything said that doesn’t agree with your current plan of attack. Not worth the effort.
Now discussing you as an interesting specimen of a particular kind of aberrant human behaviour – that’s still interesting. So I’d rather do that. It’s not aggressive, because it’s not aimed at you, I’d much rather discuss this issue with the people around here actually interested in doing so – you’ve demonstrated over what has to be nearly a hundred individual comments that you aren’t
…
@John Morales:
Read my comment again dumbass. You don’t know the difference between APM and EAPM.
Why? Do ya think re-reading it will retroactively change what you wrote?
I watch SC2 stuff, and have for half a decade.
I know whereof we speak.
(What, you think bots spam pointless clicks?)
—
Anywaty, you want to imagine that the bot was not throttled down to APM < [pro average], go ahead.
I actually watched most of those matches, after the fact. It was playing.
The bot was pseudonymous during the trial, and different agents were used for each race.
Point being, it was proof-of-concept.
And yeah, some top players worked it out, but it was never easy.
Nothing like your initial claim about huge ground armies being defeated by basic air.
—
But hey, surely the technology has not advanced one whit since 2019, right? ;)
@172 snarkhuntr
I am interested. I have some time before my workday starts. What we have is a specimen of troll in his natural habitat. The habitat is a skeptic forum, his hunting grounds. The game he is playing is “How many people I can get to clean up my bullshit before they get angry enough to give up”. You see, because most people here are skeptics the plan is actually very simple. Spread falsehoods, half-truths and assorted bullshit and wait until somebody feels the call of their civic duty to clean up the mess.
That’s like a heroin to skeptics. Defeat the falsehood, prevent the spread of bullshit, the guy who was wrong learns the truth and stops spreading falsehoods. Its the best case scenario. And that’s what might happen if the interlocutor was engaging in good faith. John is not discussing things in good faith. Because that’s not his game, his game is to waste the time of as many people as possible. While slinging insults as additional “noise” in the communication, which most people are too polite to call him on. That’s what brings him pleasure.
The only way to win is not to play. Do not clean his bullshit. But then you have a shitty space full of falsehoods and bullshit. And a mental itch you can’t scratch. And that’s why you should ban trolls the first time they piss on the bed.
@John Morales:
“(What, you think bots spam pointless clicks?)”
Holy shit. Take a break buddy. Either you are being incredibly badfaith or your brain is turned to mush because the point is that humans spam useless clicks so 250 APM for bot is 250 EAPM whereas a 400 APM for humans is much much less. What a fucking idiot.
<snicker>
You are so fucking clueless.
(You imagine you are not being passive-aggressive, best as you can? Heh)
Sure. Here, for ya:
For you, I’ll make it evident that the timeframe does not match the very same period as AlphaStar.
(Can’t really savour your chagrin if you don’t actually get it, and alas, it has come to this)
[Gotta love it when someone imagines I am bullshitting. Not the safest path to take]
I am not watching your winter content. Spell out the argument buddy.
Thank you, jo1storm, for putting it so eloquently @176. That’s a familiar pattern I’ve seen with lots of trolls. Frustratingly familiar. Same tropes, different topics.
lotharloo, your pleading pleases me.
There is no argument, mate.
Point is that Serral averages around 450apm.
Effective? Well, he’s GOATish.
Clem, Reynor, all well over 400APM.
(Effective? Well, you know, they are pros)
—
Don’t wanna watch it, don’t.
Yet, there it is.
Remember this?
Nope. It’s barely par for GMs, nevemind pros.
I used to like BeastyQT, back in the day. UThermal these days.
They sorta (in some senses) smurf but they are open and genuine about it.
(Something to which I relate)
Anyway.
Perhaps get with it, mate.
Me, earlier (@98): “A bit like restricting a chessbot to a measly IM level.”
Here’s a clue for you: https://en.wikipedia.org/wiki/FIDE_titles#International_Master_(IM)
—
Still, fine.
The throttling back meant it could not beat all the pros.
(I actually watched those matches, and how the best grokked what was going on. Did you?)
<l;snicker>
—
You two are so very cute as you cuddle against the monster.
It’s so fucking obvious you can’t dispute me, that your egos are fragile.
(I would pity you, were I that kind)
Okay John Morales, since you insist on being an idiot I’ll ask directly. What is serral’s average EAPM? If you want to compare it to 250 average EAPM, that’s a very important number to know, isn’t it?
Also, boasting that your AI reached GM using an 250 EAPM where the average GM has 200 APM is pretty silly and shows that they didn’t really solve SC2 and their results were underwhelming.
Not to mention they did not have a single AI and they hadany different variants most likely because their AI could not do diverse strategies and builds the same way a human pro can do.
John, why talk about “conceding” if you’re not obsessed with declaring victory as a means of stroking your ego?
It most certainly exceeds “22 agent actions per 5 seconds, where one agent action corresponds to a selection, an ability and a target unit or point, which counts as up to 3 actions towards the in-game APM counter”.
Fucking clueless.
If you compare the rating of the average International Master in chess, you are already way above the average chess rating.
GM is around 4800, Serral and the top pros exceed 7200.
These are all facts.
Heh.
Such idiocy!
What it shows is that they were competitive at the GM level in a proof-of-concept trial.
Back in 2019.
What, you imagine it was done to get AI players to $profigt$ from playing in e-sports tournaments?
(sigh)
Right.
Surely, a mystery for the ages.
—
(Psst. Wanna be introduced into the mysteries of the occult?)
The dramaturges of rhetoric?
The wherefore of the whichness of the why?
For you: https://www.thoughtco.com/what-is-concession-rhetoric-1689901
[arcane esoterica FTW!]
What’s funny is that I’m literally chortling with joy.
Fun is to be had, and funny you are.
(More, more!)
@John Morales:
Citation “Trust me bro”. That a lot of words for declaring that you don’t know what is the average human pro’s EAPM. Well done troll.
Gotham Chess is a good channel, but over the last couple of years it’s grated a bit on me.
Anyway, Levy is an IM and has on and off tried to get to GM.
He would make a mockery of you and I, lotharloo. Which is the point.
Doesn’t matter what the (abstract type) game may be, now in the third decade of the twentieth century they can be iteratively made into superior players to humans.
(The extrapolation is obvious, or at least it should be)
<snicker>
Feebly done, would-be troll.
—
But hey, prove me wrong.
Serral. Clem. Reynor. Maru. Dark.
Wanna argue that they fail to meet that standard?
(heh)
Nobody has done any studies on average human EAPM as far as I know and thus the alphastar versus pros exhibition matches were not convincing and they were underwhelming. SC2 is designed to be an exciting and challenging game for humans with many timings set to be difficult to react to but not impossible for humans. Therefore, deciding what would be fair conditions for such a match is non-trivial and I definitely don’t want to delegate it to the company that builds the AI.
I mean, it should be fucking obvious, even to such a jaundiced person as you, lotharloo.
Nowhere near as fast as the best, but faster than most.
You can’t dispute that, but it blows away your attempted insinuation that it was just seeing all the map and spamming actions that won the day. And your bullshit claim about not being able to deal with air.
Anyway.
Surely it is you who has shown me up, rather than the other way around.
No? ;)
Right. Whelmed, you are not.
Nevermind.
So, here is a list: https://liquipedia.net/starcraft2/Players_(Korea)
Of that set of players (only Korea, remember), what proportion do you in your esteemable estimation esteem are below that feebler APM to which AlphaStar was restricted?
I love how you ignored that it only “saw” the map and had to scroll at the usual speed and move the “cursor” at the usual (max) speed and so forth, BTW, but you probably don’t want me to elaborate.
I mean, I can, if your plea pleases me, but you would not really like the result,
FFS. What part of “proof of concept” is obscure to you?
Bah.
BTW, that’s a real match to which I linked.
I put it to you that the reason you don’t care to watch the gameplay is because it will show me to be 100% correct.
(If only you’d got the IM to GM (nevermind super-GM) thingy, you would not look so lost)
Maybe pulling opinions out of your ass and presenting them as solid arguments are normal for you but I am not interested in picking an arbitrary number and declaring it to be a “reasonable estimate for serral’s EAPM”.
About the scroll thingie, the AI played one map on that setting which it lost, as far as I remember.
So cute, you are!
Clueless, but cute.
So.
Serral is way faster than that restricted version back in 2019, but others are faster.
He’s just GOATish, that’s all.
I’ve already provided existence proof (#179).
The video you refuse to look at, since they are both ridiculously fast players.
Who also way exceed the AI’s restrictions.
(Did you imagine I knew not whereof I spoke? Heh)
FWIW: https://news.blizzard.com/en-us/starcraft2/22933138/deepmind-research-on-ladder
Oh, right. It’s you.
#202 referred to “About the scroll thingie, the AI played one map on that setting which it lost, as far as I remember.”
It played one season as all three races.
All the maps other players had not vetoed.
It was ladder play. Anonymously.
People like John are why I’m with PZ on the topic of debates. Debates are not useful. Lot of self-rationalizing adversarial attitudes and perverse incentives to “win” get in the way of doing anything productive. That’s why they’re boring and disappointing to the point of frustration.
I’d rather have a conversation that introduces me to the unexpected.
“Serral is way faster than that restricted” citation, trust me bro, I know things. I know winterStarCraft, I watch videos online!
That very video (saw it at the time) random pick.
timestamp 8:03: Reynor 604 APM Clem 436.
Both go higher, at times.
Now, you can argue all you want that’s nothing like their EAPM, and sure.
But you know, they are pros. Basically, how they earn their living.
—
You know, one of us is full of shit.
(See? We concur, no? ;) )
Yeah, reynor’s EAPM is way lower than 10 actions per second. How much lower? We don’t know. Would have been nice if you could acknowledge this and move on and not try to win every argument on the internet like a fucking loser.
What a weird thing to say.
(You know the very name of this blog? It’s not ‘follow the leader)
Heh. So clueless!
(You imagine I’m debating?)
Heh. You imagine I am a trope identifier of some sort?
(Nah, of course, but what else can you come up with?)
Ah well.
So bored and so disappointed are you, you have to note that.
(Which, strangely, ineluctably means it’s remarkable; else, why the remark?)
Irony is dead.
But sure, you expected me.
You are most evidently not having a conversation that introduces you to the unexpected.
Right?
:)
—
Um, that video is Clem vs Reynor.
And yes, Serral is indeed faster than that restricted version of the research AI.
(The videos are online, if you disbelieve me)
Some people are so clueless!
10 actions per second is 600 actions per minute.
(math)
BTW, you do get you’re getting excited over a gamebot instead of a chatbot, no?
One done in 2019 as a proof-of-concept, and which still beat 98+% of human players.
That is what you wrote, you fucking troll:
“timestamp 8:03: Reynor 604 APM Clem 436.”
Maybe first make up for fucking mind what it is that you are arguing for.
BTW, Pig, Rotterdam, Uthermal, MaNa, Rotterdam, Neuro, etc. Not just Winter.
So many SC2 content providers out there.
&
Now, you might well imagine that pros such as Clem and Reynor (Reynor 604 APM Clem 436 APM) are just bashing their keyboards like monkeys on meth. Nevermind their winnings.
(they know what they do; it is a joy to watch a world-class player clash with another)
—
I don’t need to
, I did before I started.You wanked on about how the AI could see the whole map and how it spammed untold APM and how it was fucked by an air unit. I provided context.
I even provided the analogy of restricting a chessbot to a mere IM level, on the similar basis that it would only beat 98% or so of players. GM is a big step beyond that (Gogtham Chess if you care), and then super-GM (an unofficial rating).
Anyway.
That you did not even get I was disputing your naive and simplistic and incorrect claim about the SC2Bot(s) rather than arguing for $SOMETHING$ is quite informative.
Gotta love it. Bluster, and nothing more.
You wanna prove me wrong?
Go to that video, check the APM at that timestamp.
OK, but surely these top-5 world-class players are just spamming keystrokes and mousemoveclicks?
Must be so, because so far as you are concerned, 22 actions per 5 seconds is insanely high.
(Apparently, most pro players are insane, in your estimation)
John Morales:
NAH, I am imagining you bashing your keyboard like a monkey on meth. They do a lot of intelligent but redundant inputs and we don’t know by what factor.
Such a credible claim!
<snicker>
—
You know Walter Mitty?
(He too had an imagination)
Maybe I should revive something like the Doggerel Index so I can just post links in response to common huckster and dupe thought-stopping cliches and evasive tropes. Of course, I once had a troll accuse me of being predictable because I had multiple prepared responses to their tired cliches a few years in advance of meeting them.
Nobody gives a shit whether you are snickering, laughing or sharting but the reality is that alphazero had an impact and it completely changed how chess engines were coded but alphazero had no impact. That is how you know it was underwhelming.
RR, you know what an informal fallacy is? A fallacy of irrelevance.
(Not one of inference)
But hey, revive away all you want.
(Those were the days, eh?)
that should be alphastar.
<snicker>
Stockfish is the thing.
(Alphazero was basically a proof of concept)
Heh.
No.
That is how you know it was underwhelming.
—
You don’t even get the implication, do ya?
Look: that system (that got to GM) was not intelligent.
It follows that one need not be intelligent to get to GM.
Etc.
(It’s not that the dog’s poetry is shitty, it’s that it can do poetry at all!)
Anyway.
If nobody gave a shit whether I were snickering, laughing or sharting, I would not get a comment about whether or not anyone supposedly gave a shit whether I were snickering, laughing or sharting.
But I got one.
From someone who ostensibly claims nobody gives a shit.
(You just shat yourself, can’t really deny that, can ya?)
(Gotta love it when people imagine they know stuff, then reveal just how much stuff they know)
Stockfish switched to NN based eval function after alphazero. Alphazero had impact, alphastar had none.
@jo1storm, 176
Probably far too late into a rapidly fading topic to continue, but I do differ slightly in my opinion of John from what you’ve expressed. For certain, he’s always been seen around here as troll-ish, from what I can tell. But the trolling here somehow seems different to me.
When I see what he’s posted and the way he replies to people’s arguments, something jumps right out at me: Other than a tiny bit of financier-fellating early in the topic, John has only substantially engaged in arguments about the prowess or lack thereof of AIs playing starcraft. Any other criticisms or counterpoints to his posts are either ignored completely, handwaved away, or turned into straw-persons and dismissed without examination. He is also unreasonably hostile to any disagreement, seemingly unable to accept that people’s opinions might not entirely correspond to his.
It strikes me that this could be a sign of an actual problem. Have you encountered anyone suffering from a brain injury or dementia who confabulates? Often these are people who know, or once knew, quite a lot. These are people used to understanding and having a solid grasp of facts and argument. When their minds reach for what was always there and cant find it they experience distress and just invent something to fill the gap. Many of John’s responses seem to follow this pattern – he mentions complex topics but appears to be unable to actually discuss them, howling with rage and flinging chaff as he retreats, declaring victory.
It’s pretty clear from what little actual evidence he’s provided in this topic, that he is either unwilling or incapable of reading complex texts discussing these topics. He knows the names, but demonstrates no deeper understanding. The non sequitur asides meant to describe his, alleged, physical responses to topics are also suggestive of mental decline. A functioning adult would understand that such displays are uninteresting, unwanted and inappropriate, but those of diminished capacity often resort to gross displays of bodily functions and enjoy trying to ‘get a reaction’ with this kind of behaviour.
But starcraft…. a video game from over a two decades ago, and with a sequel nearly 15 years old. This is something his mind can still encompass. The soft embrace of nostalgia is often comforting to those of diminished capacity. This is a topic he can still actually argue about, and he does – at length.
I would certainly not diagnose any kind of mental condition. Not in person and definitely not through a text based argument, but his performance here leads me to truly wonder about his mental state. This is not the john I’ve seen commenting on other topics, or perhaps I wasn’t paying attention. I truly remember him occasionally contributing things of substance, not this weird drivel that he’s producing here.
snarkhuntr finds the snark.
(Not a pretty sight!)
Striking, I am.
<snicker>
, eh?
(‘Tis not me who thinks in those terms, is it? It’s you)
LOL.
Sure. In your Walter Mitty fantasies.
<snicker>
Chess is even older, and so is Go. And checkers. And so forth.
(You don’t even know how to diss, which is… well, pitiable)
<snort>
You do amuse.
Your opinions are worth what they are worth.
(But you sure are funny!)
—
The funniest thing is that now this thread is all about me.
Me.
Be proud, you are now an opinionator of John Morales, even though you still lack the confidence to do it other than passive-aggeresively in the third person and supposedly to another party.
Heh.
I’ve lost track of the countless times when people must need to find someone (anyone!) else who still disputes me to address them about me, as if I could not read that.
So weak!
Ah well, I can try to squeeze as much juice as I can before rot and sourness sets in:
A doppelganger, perhaps.
(Maybe the MI-Gos? Brain in cylinders, that was their thing)
Heh.
Well, one more.
(I’m sensing a bit of recalcitrance, as well as due cowering)
Way to try to diss me!
—
Right now ( I don’t need a feed, I don’t need a shit or a piss), what I would most like is for you to attempt to respond to me.
(How about it?)
But hey, lesser people whining about the decline (yup, this is me in my dotage, in my senescence) of others as if the problem was the disparity between what they could do and what they can now do, well…
I could elaborate, but surely anyone who is less than fully doltish can grok my meaning,no?
—
Anyway, snarkhuntr, since this how now become (so far as you are concerned) a thread about me and my putative merits or lack thereof, I do hope (need I add the intensifier ‘fervently”?) you will attempt to actually engage me.
(Think of it as a Boss Fight)
Come on!
Surely you can show my lack of knowledge.
Mighty of old, feeble now, right?
Come on. Show me up.
(Easy for such as you no?)
—
Not that I am somehow accusing you of cowardice, O self-nymed snarkhuntr, or taunting you so that I can enjoy myself even more.
(Perish the thought!)
Come on, you coward.
Nothing about “John is wrong about this or that because of this or that reason” from you, is there.
How does it feel to have your opinions without being able to justify them?
(Relax, but. Lesser people can’t help being lesser)
—
(Did you ever imagine you would find the snark? Heh)
—
Anyway, I get it.
Topic is now me, not anything do do with the actual OP.
Here I am.
(With baited breath)
Morales, dude, seriously. It’s got to a point where it’s beyond pathetic.
Anyway, I’ll take this most recent public meltdown completely unrelated to me as an admission that I was not uniquely labelling you a troll as you have previously claimed – but that your shitposting is in fact widely known and documented.
I continue to encourage you to stop fucking up the blog for everyone else, take a hint, and fuck off
“he’ being me.
Either unwilling or incapable, or (gasp) not bothered with things that are irrelevant to my claim.
—
Hey, you know about the https://en.wikipedia.org/wiki/Courtier%27s_reply ?
<snicker>
—
OK, enough for now.
I do look forward to my hammering-down by they who seeks the snark.
I have a lot of it; a shitload, actually.
Prepare to be shat upon, O snarkhuntr, unless you keep ignoring me.
Bah, who am kidding? Ignoring me just invites more defecations.
But they do suit you, snarkhuntr.
(Welcome to wrestling the pig; fuck the adage, no? ;)
Well then, no worries.
No more after that, is there?
A bit like being alive or dead; one can’t be beyond dead.
(And, may I say, welcome back! I missed you)
—
I am most amused by that supposed metaphor.
You tried that already, no?
Me, I am a snowflake, and I melt whenever I post more than a couple of sequential comments.
That’s your conceit, right?
And your supposed inference is that I lack self-esteem, no?
Heh.
Gotta love it.
Welcome back, O nemesis!
(Conducive to emesis, but that too is metaphoric, and I shan’t further explain it)
—
I have hardly ever seen anything as pitiful as that.
Continue away all you want, O NoisyFart.
You do get the dynamic, the routine, the process, the procedure, the narrative, all that has not changed in all the years you have O so piteously railed against me.
(Familiar territory. You surely know the score, O personal troll)
—
Gotta love how you basically only ever comment to try to diss me, O QuackyTroll.
(heh)
How many years now?
You are an epitome of disingenuity.
You are a bullshitter, Mr CacophonousBobbliness.
You are known to me.
(Again, welcome back. Same old, same old)
I did notice he was actually quite unwilling to talk about the AI technology in a thread about AI.
Recursive Rabbit@217:
Or maybe even The Way of the Kook:
https://web.archive.org/web/20071018164211/https://www.insurgent.org/~jhd/kookway.htm
Morales is a piker. So far he hasn’t accused us of all being sockpuppet accounts of one another, linked to a self-published bit of Kindle-fodder as proof of his authority/superiority (look ma, I’m a published author!!1!), or threatened to sue one of us. :)
@204:
OK, how about this? Three of the things that seem “especially human” and to be extraneous, if interesting, epiphenomena of humanity actually serve vital evolutionary functions (as their high degree of uniform conservation across divergent human bloodlines and their cultural universality indicate). Those functions all relate to being a smart social organism, and the first also to being a tool user.
First, music and dancing: they are salience signifiers and mnemonic aids. They got their start as something like work-songs, helping to coordinate group efforts and to pass on the how-tos to the newbies just learning the ropes; also as mnemonic aids to pass on recipes (important if adhering to them detoxifies the plants and mushrooms involved), important information (like how to identify local edible and poisonous forage), and group norms of various kinds. This is why music sticks in your head and bits of lyrics heard nine years ago easily recall to mind even though you don’t remember any random conversation you had nine years ago, unless it was traumatic (your boss chewing you out and firing you; an argument that led to a breakup; your mother disclosing her cancer diagnosis and how much time she had left).
Being sang instead of just spoken acts as a salience signal: remember this in the future, it’s something important. If talk is linguistic messenger RNA and miscellaneous signaling molecules, song was the original linguistic DNA, before writing took over as the main way of preserving information long-term.
Second: humor, which no one can seem to precisely define or explain, is group consensus-recruiting around a criticism; group recognition of wrongness in some form. Often with a constructive intent: the desire is for the thing criticized to improve or be improved, not for it to go away. For the latter, there is:
Third, swearing. Cussing. Profanity. These are alarm, and sometimes attack, signals, marking something for rejection and recruiting others to try to build a consensus that the thing in question needs removal. That can take the form of ostracism, encysting/internal exile (jailing), external exile, or outright destruction. One big clue is that there’s almost always something analogous to “damn you” or “go to hell” in every culture’s toolkit of profanity — like bee-sting pheromones, this marks a target for exclusion from the group, up to and including harassing it into leaving or outright obliterating it, whether it be inanimate or animate. These curses literally wish exclusion upon the referent, in this life and the next. Most other curses, when used vehemently, serve analogous functions, marking something as irredeemably bad in most cases. Cursing can be used as a shield as well as a sword: “God help us”, “sweet Jesus!”, and other such exclamations may alert to a danger and the need for collective action to defend against it. They may also be used on oneself to goad oneself to precipitate action. In Dante’s Peak the “sweet Jesus!” accompanies resolve to throw normal driving safety rules out the window and violate the traffic norms, because the situation is now such that doing so is less safe than not doing so. In 2012 “Get to the fucking car!” is used to get a group of people to switch tactics from “freeze” to “flight”, in a situation where “fight” was not an option. Since curse words are used for severe situations (alarm bells and war cries) they also double as intensifiers, verbal splashes of color and exclamation marks that again can signal something as important or urgent (as with the “Get to the fucking car!” example, which was too urgent to debate). Profanity can motivate people to precipitous action instead of caution or just ignoring a thing, from avoiding a hazard, to ganging up to expel an ill-mannered member of the tribe, to preparing to fight an actual war. But mostly it gets used to “expel” the object you stubbed your toe on, or unpleasant and ill-timed weather, or an irritating defect in a machine you need to use for your job. (Blast that fucking Windows 10! Goddamn taskbar froze again for the love of Christ! I’ve half a mind to blow the fucking thing away and install Ubuntu!)
RareRabbit:
<snicker>
You know how many threads there have been?
I kinda shot my wad early on.
Clueless people, such as you, don’t get that.
But hey, instead of this passive-aggressive talking about me in the third person, perhaps you could actually quote something and try to attempt to dispute that.
(As if!)
—
—
<snicker>
Go, baby!
Could hardly get more self-referential than that.
(Heh)
—
Let’s play with the dummy:
Can’t dispute that, to such as you, they may so appear to be.
Heh.
Heh heh heh.
What it might be like to be one of those saps upon whom I amuse myself, I cannot say.
(Never been that feeble)
[gotta love poseurs, no? So very cute!]
I particularly like how the remoras address each other about me, as if I weren’t here, as if that was a comfy blankie.
(Here I am. Dance, puppets!)
Heh.
Endless joy.
[from the Deepeties’R us department]
And words can be used as a shield or as a sword or as a curse.
…
And a curse can be a word and be used as a shield.
—
Heh.
Here is an exemplar of the sort of specimen that imagines they are somehow more competent than an LLM.
(Surely, in such minds as they muster, they are indeed!)
Heh.
I get this a lot.
I chat with a few people, and before you know it, they start talking to each other about me, as though I were not here,
(So feeble!)
—
Hey, losers: here I am.
Only stupid people would write that, absent actual relevance.
See, if you explicitly claim something
, that is exactly semantically equivalent to writing that it “can’t”.(Tricky, surely, for such as you)
Bekenstein Bound, let’s be plain; what specific claim that I have made do you dispute, and on what basis?
Not exactly the weirdest request anyone has e’er made, is it? ;)
Come on.
Can you but manage a credible impression of someone who actually knows how to argue, and I promise I shall reciprocate.
Go on!
↑
[heh]
No, it isn’t. Saying:
implies that in some circumstances, profanity will do that. Substituting “can’t for “can” would mean it will never do so. You repeatedly make claims parallel to this, and they are absolute tosh.
Incidentally, John, nine successive comments is weird even for you.
John Morales’ posts are now literally spam. The only things missing are random full capitalized sentences and bright background colors.
Heh.
It’s not an entailment, is it?
Other than
, of course.Almost.
You repeatedly claim I write tosh, then you always fail to sustain that claim.
(Bluster does not pass muster)
—
See what I mean?
Assert, assert, assert.
(But never ever engage engage attempt to sustain)
—
Heh.
[meta]
As happens, this thread is all about me by now.
What I say, how I supposedly think, and so froth.
(That was not a typo, BTW)
Have I ever mentioned how amusing I find these sort of stupid cargo-cult phatic phrase passings?
(It is so idiotic!)
Almost like an AI, no?
Just a robotic recapitulation of catchphrases.
—
Still. Gotta at least have some inkling of what’s what to be embarrassed by that.
“Incidentally, John, nine successive comments is weird even for you.”
I keep hoping against hope that one of the cowards will attempt to actually engage me.
(Prodding, I am; otherwise dead, this thread is)
[yoda-speak, I do]
Being talked about in the third person, when evidently I am present, is a sort of psychopathy, I reckon.
Here I am, I shout.
For nine consecutive comments, each spaced.
—
So.
Kudos.
At least you address me, rather than the supposed audience.
(Coward, you are not. And you know I bullshit not)
BTW regarding my alleged spamming: here is the origin of the term.
(Spamming the original spam is kinda homage, no?).
So.
https://cs.stanford.edu/people/eroberts/cs181/projects/spam/sketch.HTML
— begin script quote —
Scene: A cafe. One table is occupied by a group of Vikings with horned helmets on. A man and his wife enter.
Man (Eric Idle): You sit here, dear.
Wife (Graham Chapman in drag): All right.
Man (to Waitress): Morning!
Waitress (Terry Jones, in drag as a bit of a rat-bag): Morning!
Man: Well, what’ve you got?
Waitress: Well, there’s egg and bacon; egg sausage and bacon; egg and spam; egg bacon and spam; egg bacon sausage and spam; spam bacon sausage and spam; spam egg spam spam bacon and spam; spam sausage spam spam bacon spam tomato and spam;
Vikings (starting to chant): Spam spam spam spam…
Waitress: …spam spam spam egg and spam; spam spam spam spam spam spam baked beans spam spam spam…
Vikings (singing): Spam! Lovely spam! Lovely spam!
Waitress: …or Lobster Thermidor a Crevette with a mornay sauce served in a Provencale manner with shallots and aubergines garnished with truffle pate, brandy and with a fried egg on top and spam.
Wife: Have you got anything without spam?
Waitress: Well, there’s spam egg sausage and spam, that’s not got much spam in it.
Wife: I don’t want ANY spam!
Man: Why can’t she have egg bacon spam and sausage?
Wife: THAT’S got spam in it!
Man: Hasn’t got as much spam in it as spam egg sausage and spam, has it?
Vikings: Spam spam spam spam (crescendo through next few lines)
Wife: Could you do the egg bacon spam and sausage without the spam then?
Waitress: Urgghh!
Wife: What do you mean ‘Urgghh’? I don’t like spam!
Vikings: Lovely spam! Wonderful spam!)
Waitress: Shut up!
Vikings: Lovely spam! Wonderful spam!
Waitress: Shut up! (Vikings stop) Bloody Vikings! You can’t have egg bacon spam and sausage without the spam.
Wife (shrieks): I don’t like spam!
Man: Sshh, dear, don’t cause a fuss. I’ll have your spam. I love it. I’m having spam spam spam spam spam spam spam beaked beans spam spam spam and spam!
Vikings (singing): Spam spam spam spam. Lovely spam! Wonderful spam!
Waitress: Shut up!! Baked beans are off.
Man: Well could I have her spam instead of the baked beans then?
Waitress: You mean spam spam spam spam spam spam… (but it is too late and the Vikings drown her words)
Vikings (singing elaborately): Spam spam spam spam. Lovely spam! Wonderful spam! Spam spa-a-a-a-a-am spam spa-a-a-a-a-am spam. Lovely spam! Lovely spam! Lovely spam! Lovely spam! Lovely spam! Spam spam spam spam!
.
Transcribed 9/17/87 from “Monty Python’s Previous Record” by Jonathan Partington ([email protected])
Quite literally, they are not.
Not edible, not a joke.
You, on the other hand, are amusing.
(That’s what lothar left in the loo, obs)
Bekenstein Bound @237
Oh, those are some neat ideas. I might have to look into that.
John, you’re willing to name drop AI stuff, but you’re not willing to get into the nitty-gritty of how it works. That’s why you demanded payment for doing the thing you can’t be bothered to do: research scholarly articles, even though it would settle the issue much more quickly than your obvious posturing. It’s a desperate attempt to reinforce your bluff after we’ve called it.
Whatever the case is with your mental health, it’s become clear to me you’re desperately trying to establish some kind of sense of dominance, rather than genuinely talk about AI. It’s probably why you had that juvenile obsession with labeling me a neophyte, as if being around longer is an iron-clad guarantee of being better at it.
Involute Bunny, you are exceedingly cute.
(I do like the floppy ears)
What, did someone supposedly ask me that, hitherto?
(Wanna quote that?)
Ask the stormy Jo, apparently his employer spent tens of thousands of dollars on training him on Azure.
(That’s AI shit, and he is most ungrateful for his bounty)
—
It exceeds yours by far.
(Such wankery!)
But sure, since you ask, here’s the third video on a series I watched a year ago:
—
<snicker>
You never got that was but rhetorical mockery?
heh.
Nah. I was playing with you, because you are such a sap.
So crunchy, you are.
Oh yeah.
Desperate, that’s me.
Heh.
You do you, O caperer.
(And you are so very cute when you do it!)
[I do like the feisty ones, but they never last]
[OT]
Oh, sorry bunny.
(not really, but hey. Gotta make that clear to such as you, which is a bit of a shame, and need not be).
If you refer to lotharloo’s (flushed, that one is, down the loo) claims that one type of AI is not another type of AI, where do you imagine I disputed that, O Bunny who recurses?
—
Getting the vibe yet, O floppy-eared one?
For you, this once, since I still have some faint hope you’re not one of those who get all hung up about me.
Be reasonable, I shall be reasonable in return.
Don’t believe me? Perhaps just try it.
(The converse applies)
—
Anyway.
What part of “(Point was not about elapsed duration, but about the level of competence)” was obscure to you?
Me:
John:
Level of competence was baked into what I was saying: Being around longer is not a guarantee of being more competent.
Also, I wanted scholarly articles, not a YouTube video. Especially not one about a technology (LLMs) that’s already peaked and has still has so many problems.
In other words, you specifically chose to troll, rather than correct the problem.
John Morales,
There’s a reason for that, John.
I pointed out @247 why your claim @244 is tosh – claiming X “can” bring about Y has one implication, while claiming X “can’t” bring about Y has a different, and incompatible implication, therefore the two are not semantically equivalent – and you have not even attempted a refutation.
One other thing I noticed about John is the obsession with seeking out quotes of previous discussion in this thread. While it’s valuable in some cases, this seems more about maintaining a blame game environment and keeping the conversation mired in the past, rather than moving on and improving the signal to noise ratio.
I think he needs to maintain a grudge, rather than back off the hostility and mockery and be willing to start again clean, like I was asking @69.
Let’s play along and use some quotes:
This train of thought demonstrates John imagining a rather silly analogy instead of the one I was actually making, one even more clueless than I thought possible at the time. There are AI technologies that can be useful, unlike crypto and NFTs, but they’re not the ones getting so much hype. I find the use of those high-profile AIs dubious, and citing how much big investors are throwing at it is certainly reminiscent of the other recent failed flops. Appealing to FOMO and apparent authority of business people is one of the things that raises red flags in my BS detector. This is why I’m asking for stuff more credible than promotional press releases, such as scholarly articles. I recognize I have a prejudice because of all the flops and failures I’ve seen, both within AI, as well as in other techbro flops like blockchain, the Metaverse, and so on, and I want information coming from sources that use the scientific method and other techniques to cut down on deception and hype.
It’s about pattern recognition. I’ve seen a lot of hucksters and dupes in my time, and they follow certain patterns of behavior. That’s why I bring up huckster and dupe tropes. The hype surrounding a lot of these big promises in AI follows that pattern. jo1storm did a much better job of breaking the pattern @136 and @145, but as he acknowledged, for every worthwhile project, there’s a lot more trying to torture LLMs into doing something they inherently aren’t designed to do.
Another issue: Yes, people are investing out of a profit motive. Motives, like intentions, aren’t magic. They don’t mean their efforts are rationally going to lead to profit. What I see is the least useful AIs getting the most funding because they’re making big promises and fool people through cognitive biases.
I’m reminded of some game developers who had promising ideas they wanted to work on and were excited about, but investors were always changing the subject to metaverse integration. The investors had dreams of being able to sell artificially scarce NFT products to gullible players than making games players would actually buy. Same with live service games over single-player ones. A good live service game could get a company a steady income, but the market was saturated and monetization pushed too far to attract new players. Just because someone’s got big ambitions to make profit doesn’t mean they know how to make a profit.
@Recursiverabbit, 264,265
The argument from financial authority fails on a number of fronts, but to my mind the most fatal is on an unstated assumption. Morales was suggesting that there must be some substance to this commercial AI (talking mainly about OpenAI, I think, since that was the subject of the original post) because of all the money being poured into it. The unstated assumption is that the money is being spent on this because the people spending it think that the technology itself is going to be valuable or successful.
There’s no reason to assume that the VCs and other institutional investors putting money and compute cycles into the current crop of generative AIs actually expect that the technology will live up to its stated goals – it doesn’t have to. Lots of people made money on the dot-com bubble, and on the real-estate bubble, and on every other fad and gold rush throughout history, no matter how spurious. The lesson the smart investors learned was how to time the pump-and-dump so as to get out just before the peak, while there’s still a large army of greater fools and bagholders willing to be parted from their money.
As the old saying goes: the stock market can stay irrational longer than you can stay solvent. See also: Tesla, Uber, etc.
What’s interesting here as well is the way that Morales always insists that everyone must be arguing from the specific to the general. IE, in his view any criticism of Generative AI/LLMs, OpenAI, etc must be a blanket denunciation of all AI projects. While it’s tempting to believe that this is deliberate strawmanning, I think it just stems from his inability to understand nuance or complex subjects. There are several examples in the thread of him attempting to foist some such general argument on people who disagreed with him about specific/individual projects.
@Berkenstein Bound 237,
Also mentioning the emotional power of song – there’s a reason many armies historically went to battle with some kind of musical accompaniment. And while work songs can help coordinate effort, they also provide something of a buffer against discomfort. From my own experience, singing jogging/marching songs while on military exercises actually made the pain/effort of the exercise less intense. I would imagine that sailors working while singing a shanty felt the same about the cold wet winds.
And the group bonding – people feel closer when they sing together. Everything from Football songs to Karaoke with Hymmns, chants and other rituals borrowing from this as well. You feel closer to the members of your team/tribe/church/fandom/army when you sing in unison with them. It’s an incredibly human thing.
(An unrelated aside about song) – one of the most wholesome things I’ve seen on youtube resulted from fans of a woman, a professional opera singer and self-described vocal nerd being asked to review the absolutely unique death metal singing of Will Ramos (band, Lorna Shore).
I think people wanted her to be horrified by his demented howling, growling and squealing, but it resulted in a friendship of sorts and joint research projects where they put a camera down his throat to see where he makes those sounds. I learned so much interesting stuff from her video when she starts adorably geeking out about the structures in the throat that allow him to make such unusual noises. There’s something wonderful about watching someone with an incredibly deep passion and understanding of a topic excitedly discussing their subject. Video here if anyone wants it – https://www.youtube.com/watch?v=HZaVXE5UqaQ )
[in no particular order for this zombie thread]
How am I supposed to correct you?
I respond to your comments, and you call it trolling.
(Has it ever occurred to you I can hardly do that if there are no comments to which to respond?)
Yeah, well, everyone can see how competent you are.
—
Sure. Monkey status games.
Why should I appease your need?
You want to believe it’s tosh, fine.
One thing John has noticed about you is that you can’t keep track of the previous discussion in this thread without frequent recapitulations. Tedious, but that’s how some people are.
He thinks this business of psychoanalyzing me is a very silly endeavour.
(You remember the Fundamental Attribution Error?)
—
What a remarkably stupid thing to claim.
From the OP: “I said nothing about AI! You may praise me for my restraint.”
Be aware there have been a number of AI-related threads, and PZ quite obviously thinks the entire thing is bullshit.
As did most commenters, until I contributed a bit of reality.
Now at least some people are talking about use cases, as opposed to just dissing the entire thing.
Nobody claimed that but you. Again, you’re the one fighting against straw dummy and now declaring victory about it. My claim was always (and so was PZ’s) that it is mostly bullshit that needs to be dissed. In fact, I repeatedly claimed that it is 95% and have been waiting for days for you to call me out on that number. Instead of actually researching that claim of mine, you kept repeating 100% all-bullshit-all-the-time straw claim you invented.
The truth is, you could have caught me with 10 seconds of google searching. The 95% is my personal experience and my personal bias. The real number, as estimated by venture capitalists and investors, is 85% failed AI companies in next 3 years. Its not 1 in 20, it is 3 in 20 that are not bullshit. 1.5 in 10. If you actually cared about the issue and argued in good faith, you would have known that.
https://www.thestreet.com/technology/ai-85-pct-of-startups-will-be-out-of-business-in-3-years-major-investor-says
Btw, 9 comments in a row? Really? Are you alright, John?
Cargo-cult argumentation is your forte, jo1storm.
“The stochastic parrot has run its course.”
“As best I can tell, AI is still a failure.”
“I remember something I saw a few weeks ago about AI and how it’s a fraud.”
Was much more of it in the recent previous threads.
(Don’t you remember them?)
Look at the timestamps.
Besides, if I put all that stuff in but one comment, there would be just as much stuff, no?
Still. Your concern is duly noted.
[my sifu used that technique, I appreciate it now.
multiple comments in a row are much harder to ignore than one great big comment, and one can focus on a particular aspect]
Unless something big changes about them, I think LLM (The stochastic parrots) have likely peaked, so there’s no reason to get excited about them.
@#271 Actually, I ignored all of your comments except the last one and just read comments of other people about your comments.
This:
doesn’t equate to
As you failed to acknowledge multiple times, it doesn’t need ALL of the thing to be bullshit for a thing (in this case AI) to be a fraud or over-hyped bubble or a failure. Just MOST of it. If a thing fails 95% of the time or 85% of the time, it is still a failure.
So it wasn’t an allusion to Monty Python’s dead parrot sketch?
Point being, LLMs are a class of system, not applications in themselves.
(A bit like saying TCP/IP has run its course)
e.g. https://www.growthloop.com/university/article/llm
You are so cute! Like a puppy that wants to play.
You (ironic given your work) are conflating AI (the automation technology) with specific AI applications.
(Azure — blue sky)
Singing as recruiting more workers to a job — yeah, that works too, and as bonding agent. With humor and swearing also having bonding aspects, but increasingly usable in an ingroup vs. outgroup context specifically.
I wonder if it started as pure bonding, group singing with no real “data content”, in an early hominin, then got additional functions layered onto it as part of our acquisition of language. Are there any other extant apes that sing, for loose enough definitions of singing?
So, John, how are LLMs going to overcome those heavy limitations and challenges mentioned in your link? Somehow, I don’t think more data and better curation is going to improve it enough to justify the expenses.
What part of building systems out of components, of which some may not be LLM-based, was confusing to you?
Again: Cyc.
Handwriting recognition; image recognition, math engine, etc.
The future is yet to be written.
But remember the ostensible topic is AI.
Automating what used to need intelligence, much as power tools automated what used to take muscle, and so forth.
Now, it may be that most people think LLM when they think AI.
Do you?
@278 Again, around 95% of usages is NOT this.
There will be very little of THOSE usages yet to be written because THOSE above usages are already mostly spent. They are a solved problem, especially handwriting recognition and OCR.
Because that’s what is being marketed and what is being invested in, 95% of the time. LLMs are a technology marketed as “it would change the world” or “AI”, not other things. When an investor or layman hear AI, they hear LLM because those peddling “AI” are marketing LLMs.
There’s no irony or conflation here and this is another of your straw arguments. Its not like Geminy isn’t a massive frickin’ failure and Google messed up their search when they put it in production for customers to use. As well as plenty of other non-Azure AI implementations and specific AI applications.
You should have really read the link that I have provided to you. Here it is again
https://www.thestreet.com/technology/ai-85-pct-of-startups-will-be-out-of-business-in-3-years-major-investor-says
He repeats your claims that “Automation will change every and all industries” and when asked how, he says two things, one unhinged speculation (“We’re going to be replacing construction workers with robots that can work 24 hours a day. So I think AI is going to be a massive shift in terms of how we operate as a society.”) the other is what his is really doing (“Next Round Capital Partners invests in technology and AI startups. The firm has invested in Uniphore, a conversational-AI startup that develops virtual customer-service agents for call centers.”)
A fucking chat-bot. Replacing manual labor is what he is hoping for but what he is really investing is frickin’ chatbots. That explains your dissonance in this thread quite nicely too. “World-changing applications are gonna come, you’ll see!” vs everyone else telling you “They might come or they might not, but right now 95% of those applications are chatbots and LLMs used as customer support chatbots”).
And even that VC guy, that invests millions and hundreds of millions in AI startups, admits that it is basically guesswork.
And this is what we have been telling you all along, that most of them value things as “How much money has already been invested in this? Who is the company’s founder? Does he have name known in the industry?”
Btw, those incumbents? Google, Amazon and MIcrosoft, guys selling shovels during gold rush.
[personal]
Remember this?
Why do you ask questions of a moron?
(Back in the day, someone would have quickly appraised you that ‘moron’ is a deprecated term around here)
—
Telling me I’m a clueless trolling moron, asking me (however facetiously)
, well.Et cetera.
Not our first rodeo, Recursive Rabbit still has the benefit of the doubt, oscillations aside.
You no longer do.
See how it goes in the future.
@281
I need to understand your actual thought processes, such as they are, to efficiently counter misinformation you’re spewing so others, slightly smarter people than you are, don’t get defrauded. You are still repeating corporate AI propaganda and if innocent people get caught in that bubble, the crash is going to be worse.
Why? Because I am trying to protect those innocents.
Ah, yes.
My moronic cognitive hazard spewings.
My trollish responses to people who go out of their way to, say, call me a moron.
My imbecilic (close enough, no?) avoidance of taking it easy on you.
—
You know how bloody obvious the dick-waving we’re doing is to anyone watching?
(Monkeys posturing)
—
There is a way out of this, but it’s up to you.
You keep trying what you are trying, I will keep responding the way I respond.
—
Notice the diff between you and RR?
(So far. Prejudiced, I am not. But postjudiced, oh yeah)
—
Anyway.
You do get how idiotic and forced this little dropping is, don’t ya?
That’s the innocents who will get misled by a moron.
Who are being trolled.
By a moron.
—
(Are you proud of your efforts?)
@282
Not a moron but corporate propaganda he is repeating. In short, it seems like you’re repeating propaganda and increasing hype because you’re under the wrongful impression that you are fighting group-think and curing people’s misconceptions about AI. If I can persuade you that what you perceive as misconceptions is actual reality of market for AI as it currently stands, it is conceivable that you would stop. Not bloody likely, I admit, but possibility exists.
@282
Are you? Do you think you are actually doing good in the world? Do you honestly think that you are injecting dose of reality in the conversation and not dose of hype and propaganda? Do you honestly think that people would be more likely to listen to your arguments if you piss them off and insult them?
Is that what you’ve been doing? Tells a lot about you, really. People called you out for that behavior, maybe you should listen to them.
Well, being only a moron, I still know that’s how it seems to you.
I mean, God forbid a cretin such as I could possibly be able to think for themselves, right?
That would just be inconceivable.
Surely it takes an intellect of your caliber and your erudition to realise that I am repeating propaganda and increasing hype, despite the actual words that I have written down.
Here I thought (in my dim, inchoate way) that we were on a platform titled “Freethought Blogs” where the whole conceit is that freethinking is a good thing.
—
You are a feisty one, but I am a simpleton.
Heh.
Why you persist with this little narrative is kinda informative.
And, well, you know… kinda revealing.
(Irony lives!)
Gotta say, it takes a special kind of intellect to conclude that I am a moron who has to be understood so that my moronic misinformation which is nothing but hype and propaganda can be stymied before slightly smarter people than I are misled.
(This is one of those timestamp moments, wherein I bide with baited breath).
Your last response, O puppy, was 3 minutes after my comment, to my comment.
(I know, I know. Revealing some of my moronic technique)
—
Anyway.
Don’t do that around here, is my moronic advice.
It’s all sorts of wrong.
@ 280 [inexplicably] Resident Troll
Dude, remember the dungeon? Why are you pretending to be unaware that trolling is a “deprecated” behaviour around here?
Remember when you exiled yourself from Pharyngula for a period of years because you knew you were so utterly reviled by everyone, it was that or be banned? What changed? Did you think we’d forgotten what a shit you were and suddenly wouldn’t notice? Or what?
Ah, the old relationship; my self-appointed Nemesis, who alleges that I have melted down (quite some time ago now, but hey) and lack self-esteem.
Heh. Welcome back, SputumBlob.
Dude, to what alleged trolling do you intend to refer, other than yours to me?
No.
The issue at hand is calling someone a moron.
Go on.
Tell me it’s fine around here.
(A straight face would be best, for deadpan effect)
[Idiot. Stupid. Imbecile]
That sort of thing, Bogiferous one.
Once was deprecated.
Now, some might call this moronic:
See, there are two monkeys at it, but only one is posturing.
The other, being wise, speaks not, hears not, sees not.
Right?
You are the wise monkey, jo1storm, as exemplified by that trinity.
You know the concept of https://en.wikipedia.org/wiki/Reflective_programming ?
—
<clickety-click>
Burns original
O wad some Pow’r the giftie gie us
To see oursels as ithers see us!
It wad frae mony a blunder free us,
An’ foolish notion:
What airs in dress an’ gait wad lea’e us,
An’ ev’n devotion!
@290
Dude, people have been calling you out for your posturing and insults since comment 63. You have been insulting people and posturing since comment 15. Why are you like this?! Why are you unable to have civil discussion in good faith? Why do you consider every discussion a pissing contest you have to win?
Are you self aware enough to realize that’s an issue you have?
<snicker>
I’m a moron, remember?
(One you are supposedly trying to understand)
Quite sure it’s an issue you have? ;)
Well, being only a moron, I reckon it takes more than one person pissing for it to actually be a pissing contest.
(You do amuse)
@294
Snicker to you too. You have been snickering so much in this thread that you might as well be a horse.
Are you so thin-skinned that you can fling insults but the moment somebody insults you, you focus on it for 10+ posts and counting? I guess that you can dish it out but can’t take it.
Correct. There’s no pissing contest here, you thinking that every discussion is a pissing contest and acting like it is when it is not is YOUR problem.
I was actually seriously asking you this question: Why are you unable to have civil discussion in good faith?
Get a fucking life John Morales, you are still spamming this thread, Jesus fucking Christ.
No, lotharloo.
I am responding to people’s comments to or about me.
Is it not obvious as could be I can’t respond to comments that aren’t made?
There’s a reason it specifically happens to you. See if you can work it out.
I know why, KG.
People can’t help themselves.
I don’t need you to attempt a refutation. It would be nice if you would stop spouting tosh and then making false claims when I show that they are tosh, but I certainly neither need nor expect that from you.
John Morales@299,
No, that wouldn’t explain why it specifically happens to you. Have another go. Hint: it has to do with the way you, specifically, behave.
I say LLMs have likely peaked. He sends me a link describing LLMs, I ask about a portion of the text contained in it, and he accuses me of thinking LLMs are all there is.
Though the underlying code of LLMs may be very different from crypto and blockchain, the article John linked did mention they’re also very power-hungry, consuming a lot of processing power and electricity. That’s one of my problems: Is the time saved by the end user worth all the time and energy (both joules and work hours) put into maintaining the system?
KG, if it were merely me, how come it’s only a small subset of people who persistently keep addressing me over and over?
Some people find me so fascinating hey have to keep engaging me.
Plenty of threads where it does not happen.
I grant I have the same attitude as I did back in the day, but the commentariat is much different now.
—
RR,
Whether the cost-benefits are positive or not does not affect whether they do what they do.
If they don’t do what they are supposed to do, well, people will stop spending $$$ on them.
Put it this way: is the cost-benefit of a cruise liner travelling the seas positive?
Obviously no. But people like it, it’s a profitable business.
What about the Olympics? Worth it?
Not-so profitable, but how many people travel by air to get there and back?
Et cetera.
(Not to mention how much energy and time wars cost)
Waiting for the money to wake up to that. So far, they mostly seem to be running on the idea that they’ll change what they’re actually doing if you just feed them more and better text.
Yeah, people get entertainment out of those, but profit isn’t the only measure of what makes something worthwhile. LLMs have an environmental cost, just like cruise ships do.
Silentbob@287:
Unlikely. Humans rarely start to form long-term memories before they reach the age of 3.
<snicker>
jo1storm@281:
Noble. Also, preventing people from joining the bubble hastens the bubble’s end, potentially sparing yet more people.
However, should the bubble pop before the election, the world ends. History shows that the incumbent party invariably loses the White House if a recession kicks off during the campaigning period. In this case, that would mean AI bubble pops before Nov. 5 -> Trump -> Armageddon (Holocaust II and World War III very likely, especially as the single most consistent GOP plank this season is “deport 20 million people” even while they flip-flop all over the place on abortion, Project 2025, deregulating tech, etc. etc.).
The bubble must be allowed to go on at least until election night, whatever that costs some people, as the alternative will cost them far more. It’s a triage situation.
Recursive Rabbit@303:
(303! In a non-infinite thread!)
It occurs to me that these bubbles could have a common underlying thread there. What if there’s a conspiracy by the fossil fuel industry to delay peak demand by hyping up one gas-guzzling tech toy after the next? And what, in that case, will they gin up to be the third in the sequence “Cryptocurrency, AI, ?” …
Um, my name is on the Molly list, but not on the Dungeon.
(Such ignorance! Even BubblingBlob knows that)
[gotta love when people attempt to insinuate that I’m young at heart. Child-like, really]
PS
Yeah, but these are the boring times.
Back when, kilocomment threads were the norm.
—
Heh.
Nobody either remembers my adduction of https://en.wikipedia.org/wiki/Koomey%27s_law#/media/File:Koomeys_law_graph,_made_by_Koomey.jpg or even bothered to read it.
What was it, by the puppy?
(heh)
Heh heh heh.
Such acumen!
(Peter Zeihan would surely be super-impressed at your grasp on economics!)
BTW, in case some aren’t aware of it; a lot of the cryptomining and suchlike is done there power is cheapest; and by cheapest, I mean renewable. No fuel costs, only setup and maintenance.
cf. https://www.bloomberg.com/news/newsletters/2024-02-09/cheap-electricity-is-luring-chinese-bitcoin-miners-to-ethiopia