The slop huckster's dilemma:

[cwebber@social.coop]

The slop huckster's dilemma:

"AI assisted tools are powerful but they should only be used by experts to work on things they're already experts in"

AND

It's hard to build expertise without going through the hard problems yourself, and the human brain is not conditioned to go through the hard things when there is a magical lever you could just pull that mostly does it for you

[cwebber@social.coop]

Another contradiction: I also see people saying both "there are a lot of errors sure, which is why I only use it on things I'm already an expert in" AND "AI really helps me explore new fields I don't know about"

[riley@toot.cat]

@cwebber So now you know which kind of cognitive style tends to get people caught in the quicksand slop the fastest.

[pinhman@mstdn.ca]

@cwebber an expert becomes one by making every mistake, understanding every failure by experiencing it and finding the short-comings, bad assumptions, poor logic ..learning

[shanecelis@mastodon.gamedev.place]

@cwebber “In a world of bullshit jobs, we figured a bullshit generator could save us some trouble. And it looked like it did… at first.”

[modulux@node.isonomia.net]

@cwebber Speaking for myself, I have used it to teach myself maths I didn't know, and yes, there were errors. So what it requires is being extremely disciplined about evaluating every step in a proof, making sure it's valid, ideally reproduce the work oneself. (I know there are other ways to learn maths, but few which are accessible to me. But this may extend to areas where results are sufficiently formally constrained.)

[dave@alvarado.social]

@cwebber then there's the final boss: you're a lawyer or Ars Technica writer or whatever, a proven expert in your field, and you use AI without checking anyway.

[chris_evelyn@fedi.chris-evelyn.de]

@cwebber It‘s like when they say „AI will only replace junior programmers“ and I want to shake them and shout at them where they think senior programmers come from