As humanity has found yet another way to pass the buck, it’ll be interesting to see the diminishing returns of LLMs as they begin to feed more and more on derivative content made by LLMs.
It’s interesting, because people say they can only get better, but I’m not sure that’s true. What happens when most new text data is being generated by LLMs or we accidentally start labeling images created through diffusion as real. Seems like there is a potential for these models to implode.
They actually tested that, trained a model using only the outputs of the previous generation of model. It takes less iterations of that to completely lose quality than you’d think.
Even before the LLMs, back when I was on reddit I would sometimes see conversations between bots that were 3 or 4 bots replying to each other with scraped content (usually in the personal advice subs) and getting upvotes.
I only noticed because I used to hunt bots as a hobby.
As humanity has found yet another way to pass the buck, it’ll be interesting to see the diminishing returns of LLMs as they begin to feed more and more on derivative content made by LLMs.
It’s interesting, because people say they can only get better, but I’m not sure that’s true. What happens when most new text data is being generated by LLMs or we accidentally start labeling images created through diffusion as real. Seems like there is a potential for these models to implode.
They actually tested that, trained a model using only the outputs of the previous generation of model. It takes less iterations of that to completely lose quality than you’d think.
Do you have any links on that, it was something I had wanted to explore, but never had the time or money.
They go insane pretty quickly don’t they? As in it all just become a jumble.
Given that people quite frequently try and present AI generated content as real, I’d say this will be a huge problem in the future.
Microsoft has shown with Phi-2 (https://www.microsoft.com/en-us/research/blog/phi-2-the-surprising-power-of-small-language-models/) that synthetic data generation can be a great source for training data.
Even before the LLMs, back when I was on reddit I would sometimes see conversations between bots that were 3 or 4 bots replying to each other with scraped content (usually in the personal advice subs) and getting upvotes.
I only noticed because I used to hunt bots as a hobby.
It’s cat farts all the way down.