[orleyhuxwell]

I am not tired by this anti-AI stuff. As a person who uses it in very limited capacity, also as an ML/computer vision developer and researcher with 10 years of commercial experience with it, I want much more anti AI stuff.

Low quality (low precision) news, code, marketing, diagnosis, articles, books, food, entertainment (shorts, tik-tok), engineering is in my opinion the biggest problem in XXI century so far.

Low quality AI usage decisions, low quality AI marketing, retraining, placement, investments are accelerating the worst trends even more. It's like Soviet nuclear trains - just because nuclear is powerful and real it doesn't mean most of it's applications made any sense.

So as a pro-AI person and AI-builder in general, I want more anti-AI-slop content, more pro-discipline opinions.

[jfyi]

I think you have hit on something. The problem isn't the tech, it’s the eye of the user.

The same person who ignores a crooked door frame or a CSS overflow now has a "mostly right" button to bring mediocrity to scale. We unfortunately aren't invested in teaching craftsmanship as a society.

[seanmcdirmid]

Both sides are against slop, what we are arguing about is basically the position that “AI can be used for useful things” vs the “all AI is slop” positions, the latter being based on hasty generalization fallacies (some AI is slop so all AI is slop).

[orleyhuxwell]

Sorry, but I don't think so.

I think without AI the effort of producing slop code or art that sort of looks like a real thing on the first glance is let's say 5% of the effort needed for the real thing that actually works flawlessly. LLMs and diffusion models bring it down to 0.5%.

They are also really good at faking comprehension and make recognizing real experise from phoney cosplayers harder for busy managers, officials, execs, politicians etc.

So while AI CAN be used for useful things, it very rarely is and it requires more discipline than most people are willing to invest.

Also, the way AI is trained on stolen and random low quality content is deeply disturbing.

So yeah, while I'd like anti-AI rants to be more precise and nuanced, in general AI in 2026 is mostly a missuse of a technology with great potential.

[seanmcdirmid]

Both sides are against slop made by AI, not that bit sides can make slop, which is true but irrelevant.

Still so many misconceptions about AI on HN, but I guess it’s just how techies are these days.

[orleyhuxwell]

Yeah, but to me it's like arguing 'smoking can be more beneficial than harmful to your help, because you take a break and meet new people and loneliness is bad; you should just be disciplined to never smoke more than 2 cigarettes a day and another hard limit is 6 a week'.

Yes, that VERY RARELY would work, but for most (and I mean 98% not 51%) people it doesn't. It's non symmetrical.

> bit sides can make slop, which is true but irrelevant

Idk, to me avoiding slop seems like the most ethical thing a professional can do these days.