Let’s Learn How to Write, NBC


A basic mistake like this in a headline? Really?

I know grammar postings are eye-rollingly passe, but this kind of thing seems to be getting worse. Or maybe I’m finally at the “you kids get offa my lawn!” stage of life, or something. I’ve been finding a lot of really bad spelling mistakes, mostly on the “new media” but sometimes on the “old media” as well.  I suspect it’s because both are hiring the same unpaid interns, or something:

lets

Perhaps this is some of what’s going on. If the computer doesn’t flag it as language error, it’s not language error:

lets2

So, consider this not a “writing flame” but a “software and algorithms whinge.”

Many people don’t know that authenticity is under attack from yet another angle: AI journalists. Apparently most “city desk” news is produced by AIs. The AI tools take content facts and shuffle them around into mathematical patterns of narrative then randomly inject “realistic-looking” errors. The end result is indistinguishable from Milo Yiannopolous.

I’d sign up for the Automated Insights free trial, load a bunch of PZ’s postings and some Chuck Tingle and mix the DNA: a new literary-scientific genius would be born! Slammed in the butt by HOX genes!

I remember listening to a talk Rob Pike did at USENIX, about his USENET news-bots Mark V. Shaney and Bimmler. It was funny, back then, but now I realize Rob was revealing the future of political commentary. Here’s why it bothers me: tools like Automated Insights take “facts” (in the form of a spreadsheet) and expand them into “verbal nougat fluff” with embedded facts swirled in like pistachios. We could just read the facts.

I, for one, welcome my new AI commentator!

divider2

The Atlantic – “A Computer Tried And Failed To Write This Article

WIRED – “This News-writing Bot Is Now Free For Everyone

Article Spinning – how spammers do it

Automated Insights Free Trial Signup

 

Comments

  1. says

    Caine@#2:
    I had no idea.

    Well, that’s kind of the point. ;)
    It’s hard to get to experiment with it. I wanted to see how it’d do at writing a few atheoskeptical rants, templated off of PZ’s “so and so is a dishonest buffoon” structure, and then plug in a few names and “dishonest about [thing]” and see what it does.

    Trump may be a badly implemented AI rantbot. Ever think of that?

  2. says

    Chigau@#4:
    Had anyone actually seen this “Marcus” person in meat-space?

    Lots. If you search me on youtube you’ll find documented incidents of me speaking at public conferences.
    Lemme tell ya, the special effects teams go NUTS with those gigs. There was a time a couple of years ago in Malaysia where the harness holding the lizards together came apart and “I” had to keep one of my “arms” stuck in my pocket because the entire “shoulder” was falling out of the shirt we were wearing.

  3. Siobhan says

    The end result is indistinguishable from Milo Yiannopolous.

    I’m pretty sure my last outhouse visit is indistinguishable from Milo Yiannopolous.

  4. invivoMark says

    Oh, creative apostrophe use. An acquaintance from high school who is on my Facebook friends list decided that apostrophes do not serve grammatical purpose, but only serve to alert the reader that a word is about to end in an “s.” She alway’s litter’s her post’s with word’s ending in sibilant’s.

  5. Nathan says

    “The end result is indistinguishable from Milo Yiannopolous.” LOL.

    There’s also an increasing number of news headlines that make absolutely zero sense. I can’t think of any right now, but that pisses me off as well.