ASCII Tweezers

if you thought this was written by AI

You find yourself at the beginning of an article. It's a place you've been countless times before, and probably will be again a few more times today. You pick up different articles for different reasons but one thing is certain: if you thought this was written by AI, you would stop reading on the spot. There's already too much to read, more than a lifetime's worth. The last thing you need is to get tricked into reading some low effort slop.

And you have been reading for most of your lifetime so far. Growing up, learning to read was liberating and exposed you to all sorts of information, diverse and confusing and interesting at a scale your assorted caregivers couldn't possibly have replicated orally. That was a long time ago.

You've always had some basic preferences about what you read, but over time your sense of taste has shifted and been refined. A vague awareness of quality emerged, difficult to pin down yet increasingly visible to your oh-so-discerning eyes. Even if you can't articulate where quality comes from, it finishes in just a handful of ways: in pleasure, utility, or ego.

Really, it's hard to disentangle your personal taste from its social aspects. Some authors you like are considered a bit trashy by others, and you know this, and it's not as if you aren't allowed to read or enjoy them, but it does feel necessary to make the requisite acknowledgements, to at least signal your self awareness to your fellow subway riders.

Now there is a new kind of trash. It hasn't entered the books you read (or so you hope), but it is becoming omnipresent in shorter form writing. And it feels important to train yourself to recognize and discard it. You think of your elders getting tricked by surreal slop montage videos of dogs rescuing children from collapsing buildings, and you shudder. That will never be you! Not at any age, not ever.

The techniques for detecting AI authorship are idiosyncratic and change frequently. There is a cat-and-mouse game where the fixations of each model become broadly known, broadly mocked and then feverishly scrubbed out of the next version. As an unfortunate side effect you find yourself nervously policing your own writing so that it avoids all of these emerging cliches. If it's embarrassing to read AI slop without realizing, then you wouldn't want to be accused of producing it. You can live without bringing up goblins at every opportunity, but mourn the innocent time when you could use a semicolon with unimpeachable self-confidence.

Well, at this point, you're most of the way through the article. You have more than enough data to make your determination.

You reflect on what you have read so far and how it made you feel. The title turned out to be clickbait, but that's not particularly rare. But the mixing of more traditional literary tone with the conversational, and the intermittent TED talk-style emotional impact phrasing strikes you as suspicious. You start to feel queasy. Looking back, this whole thing has been circular and vacuous; sentences that were vaguely satisfying by themselves have added up to nothing, no takeaway, and now you're holding the bag.

What was the point? It's almost as if the author did not read their own work, or else they would have fixed it or at least spared the world. If they don't care enough to read it, why should you? The lack of intentionality and care sickens you; it is fatal to the piece and disrespectful to you, O Reader.

There's only one thing left to be done. Prior generations of internautes had a sort of mantra for this; the words have been muddled through many retellings, but the essence remains intact. You close the tab. You unplug your router. And, with a steady hand, you dial 911.

#AI #culture #nonsense