Earlier quoted context omitted.
It’s shared context for anyone familiar with the field.
Amusimgly, also how cults and fascism works.
LLMs are real, AI is fake
41–45 of 45 posts
Re: LLMs are real, AI is fake
#42Earlier quoted context omitted.
> Err, no? That's not at all how llms work. The Transformer architecture that almost all LLMs use are composed of many layers in sequence, each containing an attention component and a neural network component. The attention component copies data between tokens/vectors in the current context, and the neural network adjusts each individual vector in the context. Notably, the neural network component behaves like a comp…
>Notably, the neural network component behaves like a compressed index of the training data. The "Compression" part is indeed about the model finding order in the training data. But this is not about applying some predefined compression algorithm on it, but by actually learning an algorithmic representation of the data. This has nothing to do with "consulting the training data", because the training data cannot be re…
Re: LLMs are real, AI is fake
#43Earlier quoted context omitted.
>Notably, the neural network component behaves like a compressed index of the training data. The "Compression" part is indeed about the model finding order in the training data. But this is not about applying some predefined compression algorithm on it, but by actually learning an algorithmic representation of the data. This has nothing to do with "consulting the training data", because the training data cannot be re…
great distinction. can we use it to freely produce and share mp3/h264 copies of any media? because it is provably impossible to reconstruct original based on that information.
But keep in mind: So far there is no way to train the models while completely avoiding memorization and only including generalization. That would be a great way to avoid any copyright issues, but all attempts I have seen so far were fairly limited.
Re: LLMs are real, AI is fake
#44Some really important and informative stuff in here—I certainly had no idea just what the nature of the prompts and tooling that produced the HuggingFace exploit were. This shows fairly clearly that (as I already suspected) this was not, remotely, an LLM "going rogue." This was humans planning poorly, not thinking of the consequences of their actions, and giving LLMs too much scope and a lousy prompt.
I wouldn't even call this "humans planning poorly", I'd call it "humans pretending to plan poorly for publicity". Weasels gonna weasel.
Re: LLMs are real, AI is fake
#45My takeaway: We should be not be concerned about AI bots' "goals", we should be concerned about the goals of the companies making them. Powerful but not sentient technology in the hands of reckless accelerationists is a plenty dangerous enough thing.
I mean, of course we should be concerned about the goals of the companies. But that's a second, separate concern and it's important not to muddy the two together into a single point. We shouldn't take any of these companies at their word and we should treat their stated intentions as suspect regardless. The agents' "goals" in the specific instance being discussed is a benchmark. But it's also the case that at no poin…
Perhaps we might ALSO decide that this new explosive is so dangerous it needs specialized laws to protect the public, but that would seem premature if all available evidence indicated that the explosion would not have occurred if the company had taken the most basic precautions.