LLMs don't do anything well by human standards. Not even the one thing (programming) they are supposed to be good at. It's disheartening how many people are prepared to accept the slop an LLM puts out just because it's fast.
replyI've seen way too much humanslop to be able to reject LLM slop outright.
replyThere are areas where the average of LLM slop already beats the average of humanslop by a sizeable margin.
Possibly, although I have to say that I get PRs now where it's clear it's a miss by the LLM and was generated by a human who did not take the time to understand the subject matter.
replyIn the past, a human producing humanslop would get a serious talking to, but nowadays it's apparently a valid excuse "well, an LLM generated it". This approach is placing a huge burden on the reviewers, much more so than in the past.
The interesting thing for me is that I have a very easy time of identifying AIslop. Have I been conditioned to simply ignore or accept humanslop? Is that the direction this is all going? That at some point we will simply accept both as a product of the human existence?
replyWe shouldn't. Slop is slop. But there are a couple of differences:
replyAI slop can be more annoying to read. I mean, some human slop is as annoying as AI slop, but not all of it.
AIs can produce slop far faster than humans, so that the majority of slop at the moment is AI slop.
The difference with 'humanslop' is that it's avoidable by training or hiring a better human. With AI it's all the same, even SOTA models are still very distinctly and obviously AI in what they produce. And even very dedicated humans have a limit and aren't capable of creating low value content nonstop, meanwhile you can set up an LLM, point it at the internet and let it start filling gaps by firehosing it in low quality slosh 24/7, completely unattended.
reply[dead]
reply