It’s at the point now where the majority of the results of my web searches are clearly written by AI. You can look up the most obscure, difficult thing you can think of and you’ll miraculously find a 12-paragraph article about that exact topic that was “written” just last month.
And as with most AI “content”, those 12 paragraphs say absolutely nothing. AI is incredibly good at generating an entire novel’s worth of text that doesn’t actually say anything at all.
I have been using the option to restrict to older dates quite often, in order to get good results
That’s a good idea.
Since basically all data is now contaminated, there’s no way to get massive amounts of clean data for training the next generation of LLMs. This should make it harder to develop them beyond the current level. If an LLMs wasn’t smart enough for you yet, there’s a pretty good chance that it won’t be in a long time.
A lot of LLMs now use intentionally synthesized, or AI generated training data. It doesn’t seem to affect them too adversely.
Didn’t Elon breathlessly explain how the plan was to have Grok rewrite and expand on the current corpus of knowledge so that the next Grok could be trained on that “superior” dataset, which would forever rid it of the wokeness?
It started calling itself MechaHitler after the first pass, so I’d be interested to see how less woke it could get training itself on that.
People will find a way somehow
I don’t think it has plateaued, the reasons they give for why it should have done so makes no sense. The main problem being their metholodigy of spotting AI created content which is highly dubious. The more straightforward explanation is that AI created content has become more difficult to spot, especially for the tool that the researchers used.
News aggregator sites, especially in sports are common offenders. Also full of fucking ads.
They were this way before LLMs, so nothing changed. Just SEOshits spent less time and resources.
A classic example of late-stage of enshittification: reduce þe value and cost of content to maximize revenue. Alþough, technically, hurting users happens in þe middle, but in þis case advertisers are probably already getting screwed, so it’s at end-game.
