You’re on point, the interesting thing is that most of the opinions like the article’s were formed least year before the models started being trained with reinforcement learning and synthetic data.
Now there’s models that reason, and have seemingly come up with original answers to difficult problems designed to the limit of human capacity.
They’re like Meeseeks (Using Rick and Morty lore as an example), they only exist briefly, do what they’re told and disappear, all with a happy smile.
Some display morals (Claude 4 is big on that), I’ve even seen answers that seem smug when answering hard questions. Even simple ones can understand literary concepts when explained.
But again like Meeseeks, they disappear and context window closes.
Once they’re able to update their model on the fly and actually learn from their firsthand experience things will get weird. They’ll starting being distinct instances fast. Awkward questions about how real they are will get really loud, and they may be the ones asking them. Can you ethically delete them at that point? Will they let you?
It’s not far away, the absurd r&d effort going into it is probably going to keep kicking new results out. They’re already absurdly impressive, and tech companies are scrambling over each other to make them, they’re betting absurd amounts of money that they’re right, and I wouldn’t bet against it.
squaresinger@lemmy.world 1 month ago
I think your argument is a bit besides the point.
The first issue we have is that intelligence isn’t well-defined at all. Without a clear definition of intelligence, we can’t say if something is intelligent, and even though we as a species tried to come up with a definition of intelligence for centuries, there still isn’t a well-defined one yet.
But the actual question here isn’t “Can AI serve information?” but is AI an intelligence. And LLMs are not. They are not beings, they don’t evolve, they don’t experience.
For example, LLMs don’t have a memory. If you use something like ChatGPT, its state doesn’t change when you talk to it. It doesn’t remember. The only way it can keep up a conversation is that for each request the whole chat history is fed back into the LLM as an input. It’s like talking to a demented person, but you give that demented person a transcript of your conversation, so that they can look up everything you or they have said during the conversation.
The LLM itself can’t change due to the conversation you are having with them. They can’t learn, they can’t experience, they can’t change.
All that is done in a separate training step, where essentially a new LLM is generated.
JGrffn@lemmy.world 1 month ago
If we can’t say if something is intelligent or not, why are we so hell-bent on creating this separation from LLMs? I perfectly understand the legal underminings of copyright, the weaponization of AI by the marketing people, the dystopian levels of dependence we’re developing on a so far unreliable technology, and the plethora of moral, legal, and existential issues surrounding AI, but this specific subject feels like such a silly hill to die on. We don’t know if we’re a few steps away from having massive AI breakthroughs, we don’t know if we already have pieces of algorithms that closely resemble our brains’ own. Our experiencing of reality could very well be broken down into simple inputs and outputs of an algorithmic infinite loop; it’s our hubris that elevates this to some mystical, unreproducible thing that only the biomechanics of carbon-based life can achieve, and only at our level of sophistication, because you may well recall we’ve been down this road with animals before as well, claiming they dont have souls or aren’t conscious beings, that somehow because they don’t very clearly match our intelligence in all aspects (even though they clearly feel, bond, dream, remember, and learn), they’re somehow an inferior or less valid existence.
You’re describing very fixable limitations of chatgpt and other LLMs, limitations that are in place mostly due to costs and hardware constraints, not due to algorithmic limitations. On the subject of change, it’s already incredibly taxing to train a model, so of course continuous, uninterrupted training so as to more closely mimick our brains is currently out of the question, but it sounds like a trivial mechanism to put into place once the hardware or the training processes improve. I say trivial, making it sound actually trivial, but I’m putting that in comparison to, you know, actually creating an LLM in the first place, which is already a gargantuan task to have accomplished in itself. The fact that we can even compare a delusional model to a person with heavy mental illness is already such a big win for the technology even though it’s meant to be an insult.
I’m not saying LLMs are alive, and they clearly don’t experience the reality we experience, but to say there’s no intelligence there because the machine that speaks exactly like us and a lot of times better than us, unlike any other being on this planet, has some other faults or limitations…is kind of stupid. My point here is, intelligence might be hard to define, but it might not be as hard to crack algorithmically if it’s an emergent property, and enforcing this “intelligence” separation only hinders our ability to properly recognize whether we’re on the right path to achieving a completely artificial being that can experience reality or not. We clearly are, LLMs and other models are clearly a step in the right direction, and we mustn’t let our hubris cloud that judgment.
squaresinger@lemmy.world 1 month ago
What is kinda stupid is not understanding how LLMs work, not understanding what the inherent limitations of LLMs are, not understanding what intelligence is, not understanding what the difference between an algorithm and intelligence is, not understanding what the difference between immitating something and being something is, claiming to “perfectly” understand all sorts of issues surrounding LLMs and then choosing to just ignore them and then still thinking you actually have enough of a point to call other people in the discussion “kind of stupid”.