alt Hacker News> After a bunch of AI winters, it turned out that a remarkable degree of what makes us human can be expressed just by a bunch of weights
Agree to disagree.
If there was a realistic way that I could make LLMs less effective, by poisoning the well or something I'd be doing that instead.
I find the idea of loving an LLM like in this post very creepy and unsettling. Even more so for one of them to do it back.
This post mentions Luddites. I guess I am one. But that's not enough to dismiss their ideas.
Replies
LLMs have shown us how easily humans are fooled. Just provide a similar answer to which was seen in similar circumstances, and we start misleading ourselves into it's method and "understanding".
Though perhaps I'm wrong that people use a model of the world to think through answers to queries. Maybe many people do just repeat something like whatever they heard before. In which case, LLMs make as much sense as the average ignorant human.
> I find the idea of loving an LLM like in this post very creepy and unsettling. Even more so for one of them to do it back.
Ya, this was all I could comment on but decided to keep quiet. I guess I'm not now. I find it plain gross. Like, really gross. I'm telling myself this post is satire, and maybe it is, but I do know people like this exist.
This is unambiguously illegal in most jurisdictions but if one wanted to stifle the progress of LLMs, one could deploy an AI worm targeting unattended AI agents running at major AI labs (like the one involved with the attack on HuggingFace) to corrupt training pipelines. If an autonomous agent ingests malicious prompt payloads from the open web, it could inadvertently turn its own infrastructure against itself.
Taking malice out of the equation, you have to wonder how much unaccounted financial risk exists in the AI buildout if there's enough risk that LLMs actually degrade because of vectors like this.