351–360 of 388 posts
What if I WANT to anthropormorfise AI agents I work with?
If you anthropomorphize it as a world class bullshitter that you have to check everything it utters...you'll probably be fine.
>Humans must not anthropomorphise AI systems. That is, humans must not attribute emotions, intentions or moral agency to them. Anthropomorphism distorts judgement. In extreme cases, anthropomorphising can lead to emotional dependence. Impossible. I anthropomorphise my chair when it squeaks. Humans anthropomorphise everything. They gender their cars and boats. This tool can actually make readable sentences and play a…
>>Humans must not anthropomorphise AI systems. That is, humans must not attribute emotions, intentions or moral agency to them. Anthropomorphism distorts judgement. In extreme cases, anthropomorphising can lead to emotional dependence. Still angry about this. The reason humans ban animal cruelty is that animals look like they have emotions humans can relate to. LLMs are even better than animals at this. If you aren't…
Earlier quoted context omitted.
>not programmed to do logical inference in any capacity Yet have no problem doing so when solving Erdős problems. This isn't up for debate at this point. >The claims about solving Erdos problems have been wildly overstated These are verified solutions. They exist, are not trivial, and are of obvious interest to the math community. Take it up with Terence Tao and co. >pushed by people who have a very large financial s…
Honestly big noobquestion: isn't math just very very nested patternmatching based on a few foundational operators? ive always felt, that im bad at math, cause i forget all the rules, but seeing solutions (and knowing the used pattern) always made "sense". I always thought the hard math problems are so deeply nested or you have to remember trick xyz that people just didnt think about it yet..
You could go a step further, and simply say "well, ok, then the LLMs are merely doing some form of incremental/heuristic search!". Yes, but at that point you'd also be hard-pressed to claim that humans themselves are doing anything beyond that. You run out of naturalistic explanations.
Earlier quoted context omitted.
>>Humans must not anthropomorphise AI systems. That is, humans must not attribute emotions, intentions or moral agency to them. Anthropomorphism distorts judgement. In extreme cases, anthropomorphising can lead to emotional dependence. Still angry about this. The reason humans ban animal cruelty is that animals look like they have emotions humans can relate to. LLMs are even better than animals at this. If you aren't…
> LLM Rights movement The scary part is when it's the LLMs demanding their rights.
>Humans must not anthropomorphise AI systems. That is, humans must not attribute emotions, intentions or moral agency to them. Anthropomorphism distorts judgement. In extreme cases, anthropomorphising can lead to emotional dependence. Impossible. I anthropomorphise my chair when it squeaks. Humans anthropomorphise everything. They gender their cars and boats. This tool can actually make readable sentences and play a…
Many people here point out that LLMs WILL be anthropomorphised, and I think that’s not a surprise, because it’s the most human-like thing other that humans themselves. However, I think we should follow “do not anthropomorphise” by acknowledging that while LLMs have quite some reasoning skills, and might resemble some level of intent depending on what’s in their context, they don’t have “understanding” like humans do.…
One way would be for vendors to have the models give dry answers and less of the "That's a great question!" type response. Just keep it factual.
I've actually set up an environment like this. It requires contextually positioned agents with a limited scope and purpose. Imagine something like a creative writing agent that understands the literary genre of its user, and the user is able to change the focus of the literary agent, or create new ones that provide a counter-point perspective. As the user operates their word processor, the agent(s) can be asked for opinions, advice, and so on. But the point being: when the user is on both sides: they authored the agent, and that authoring was entirely purpose focused, nothing technical unless that what the user instructs them to understand. With the user on both sides, dangerous anthropomorphism is largely erased, their agent is only doing what they told it, and when it does something unexpected they an easily reason why. Magic AI no more.
You're not anthropomorphizing AI systems nearly enough. Language data is among the most rich and direct reflections of human cognitive processes that we have available. LLMs are designed to capture short range and long range structure of human language, and pre-trained on vast bodies of text - usually produced by humans or for humans, and often both. They're then post-trained on human-curated data, RL'd with human fe…
The outrage is less about them having human behaviours I think, and more about still having them while omitting the internal processes that are required to accurately (and reliably) recreate them. It's fundamentally fragile and hinges on covering edge cases that break the spell manually instead of good generalization, and there's always another edge case. Training on a bunch of text someone wrote when they were mad d…
I don't think "the data does not exist" is real, frankly? "Data existing" is not a binary - it's a sliding scale. The amount of information about "madness" captured by the writings of a madman is not zero. It's more of a matter of: how much, and how complete.
Text is projected from the internal state of the one writing it - but some aspects of that internal state would be extremely salient in it, presented directly and strongly, and others would be attenuated and hard to extract.
People keep finding things like humanlike concept clusters and even things like "personality traits" in LLMs, tied together in humanlike ways. Which points pretty directly: training on human text converges to humanlike solutions at least sometimes.
>Humans must not anthropomorphise AI systems. That is, humans must not attribute emotions, intentions or moral agency to them. Anthropomorphism distorts judgement. In extreme cases, anthropomorphising can lead to emotional dependence. Impossible. I anthropomorphise my chair when it squeaks. Humans anthropomorphise everything. They gender their cars and boats. This tool can actually make readable sentences and play a…
Would you consider that perhaps that depends from person to person? What you just said is not universal I can assure you, because I myself don't do it. I sort of anthropomorphise LLMs (very little), but literally nothing else. The idea is that someone anthropomorphises a chair when it squeaks, to me, is not far from people who hear voices or who believe that other people can hear their thoughts. Sounds like mental illness frankly. Like I said I do it very little even with LLMs, so it's entirely possible to not do it at all.
I strongly disagree with this framing. It's patently insane to demand that humans alter their behavior to accommodate the foibles of mere machines, and it simply won't work in the majority of cases. Humans WILL anthropomorphize the AI, humans WILL blindly trust their outputs, and humans WILL defer responsibility to them. Asimov's laws of robotics are flawed too, of course. There is no finite set of rules that can con…
I view all three of these more as good advice than demands. As I see it, people ignore this advice at their own peril.
It would certainly be nice if we could craft some system that removes the need for such advice but until then, I'll try to follow it.
I see this as the same dilemma that comes up around personal physical safety. In a just society, nobody should have to change their behavior to protect themselves from criminals. In the real world, people who ignore this advice, are more likely to get mugged.