91–100 of 381 posts
LLMs are trained to produce outputs that humans would find reasonable. That's it. They don't do anything else. To say they have feelings and that they can think/reason is insane.
They can have empathy too. Meaning, you can describe a circumstance with many parties involved, with different motivations and conflicting goals -- and it will tell you how these humans would likely feel about each other and the situation at hand.
If I understand the central argument, which is confusingly stated, the claim is that "consciousness" is a socially-determined, subjective, descriptor for certain things that one cares about. How does this make any sense? If I care about my teddy bear enough it'll gain consciousness?
(OK, more charitably, these concepts are often murkily defined, but people react emotionally to any concrete definition because they have in their mind that something being conscious and intelligent implies that it requires moral consideration, and morality is complicated, murky, and generally emotionally driven in a way that most concrete definitions struggle to adequately describe)
Wow. I have rarely read anything that made me lose more respect for the author in a single article. And this guy is a VP at Google ? And he leads an AI research team ? His third paragraph, which is the core of his argument, is stupid . He says, "As intelligent, social entities, we decide ... which other entities we regard as having inner lives and being worthy of care." Yes, that's correct. "We do not care for others…
there might be some merit to his point though I doubt anyone considering that whales might be highly intelligent and conscious beings while they were being hunted for oil.
What the author seems to observe is that this non-evidential judgement that other beings are conscious is due to us caring about them. The brain creates this perception so that we care about them, form bonds, work together, and so on -- evolution selects for cooperation. So the question of whether we will think of AI as conscious is that of whether we will care for it like we care for another being. And, in this regard, the author points out that "intelligent agents" that appear to do this work effectively together.
My personal point of view is that it's unclear what being an AI is in order for us to even talk about it being conscious. Is it the computations happening in the GPU according to a set of weights? Is it the weights themselves? Is it the architecture? Does it include the code around it, that gives it the context and prompt, and hooks the response to a terminal, or to a client host, etc.?
Quoting from the Permutation City by Greg Egan: "Opponents replied that when you modeled a hurricane, nobody got wet. When you modeled a fusion power plant, no energy was produced. When you modeled digestion and metabolism, no nutrients were consumed – no real digestion took place. So, when you modeled the human brain, why should you expect real thought to occur?"
LLMs are trained to produce outputs that humans would find reasonable. That's it. They don't do anything else. To say they have feelings and that they can think/reason is insane.
Quoting from the Permutation City by Greg Egan: "Opponents replied that when you modeled a hurricane, nobody got wet. When you modeled a fusion power plant, no energy was produced. When you modeled digestion and metabolism, no nutrients were consumed – no real digestion took place. So, when you modeled the human brain, why should you expect real thought to occur?"
LLMs are trained to produce outputs that humans would find reasonable. That's it. They don't do anything else. To say they have feelings and that they can think/reason is insane.
Sorry, but they can think and reason. They can have empathy too. Meaning, you can describe a circumstance with many parties involved, with different motivations and conflicting goals -- and it will tell you how these humans would likely feel about each other and the situation at hand.
Earlier quoted context omitted.
It is worth noting, in the same issue they make an appology for claiming that Nvidia also makes its chips and not just design them. The Economist might be a very reputed journal but their technology and especially AI related takes are absolute garbage. I say this as a regular who still finds value in their business and economics sections. But god do they talk a lot of bs about technology.
This sounds like Gell-Mann amnesia
Earlier quoted context omitted.
Humans are just baby replicators. The civilizations are what's getting uncanny. It's good the 'just token predictors' crowd is starting to pivot and hedge though lol. 'It's just the harness!' uh huh. give it a couple more months and you'll be in here regurgitatin' Gordon Wood
Baby replication (creation, actually, I don't think anyone has ever copied a baby) is not what humans do as their only function that enables them to do anything in life. LLMs are literally token predictors. You can debate over the "just", but we haven't moved onto some new text generation technology, we're just expanding that one. I don't understand why you'd scoff at facts.
Predicting the next token is an immensely powerful thing. You can encode basically anything into a sequence of tokens, any configuration of matter. If you have a perfect next token predictor, you can predict basically anything. You can simulate an entire universe. It's Turing-complete. It sounds diminutive, but it's a general paradigm. Given everything this paradigm appears to be capable of, well, why not consciousness, really?