The whole reason they seem this way is because they’re designed by us to be very competent mimics of us.
LLMs/GenAI are absolutely not conscious. They’re just a really advanced game of word association, which cab lead them to say absolutely anything in response to the right prompts.
If there ever truly is a day where we knowingly created an actual conscious AGI, I suspect it would be locked up tighter than fort knox by whichever country’s military found it first - not interfaced onto the internet to answer questions.
I still don’t understand how it can seem this way, and the fact that so many people seem to think so feels like a massive failure of the education system to instill the most basic of critical thinking skills. Once every month or two I check in to see if an LLM can achieve a half decent 1 on 1 D&D game and it always falls horribly flat within the first minute or two.
Once every month or two I check in to see if an LLM can achieve a half decent 1 on 1 D&D game and it always falls horribly flat within the first minute or two.
Surely she would make a variety of very large bodies following a theme, use them to perform superheroic acts while pretending to be a supergenius shut-in, and then fall in love with a cyborg?
You are wrong. LLMs are indeed only about as conscious as insects, if even that. They are not sapient. However, that does not mean that they have no decision-making abilities.
My point is not that you underestimate LLMs but that you overestimate consciousness. Being conscious just means having the ability to learn. LLMs are built upon trial-and-error. They aren’t programmed, they are taught.
The current generation of AIs are nowhere near a human intellect, but every year that passes, the AIs will get more and more intelligent. One day we will live in a world where AIs have human or near-human level intelligence. And when that day comes, this staunch anti-consciousness stance will be the excuse given for the enslavement of sapient beings.
So, sure, laugh about the people who mistakenly think that word-processing means sapience. But don’t delude yourself into thinking that there is something unique about a bio-brain that means it can not have a digital equivalent. Digital sapience may not be here yet but it is most definitely on the horizon.
I think you’ve misunderstood my comment, or maybe saw the unfinished one I accidentally posted.
I am not saying that AGI, or human equivalent AI is impossible.
The fact we have brains capable of generating sapient consciousness out of a network of neuronal connections means it is possible, its just a matter of getting the secret sauce.
But I don’t think intelligence is equal to consciousness. I’m sure if you gave a spider all the world’s data and the ability to talk it’d be very coherent and could even pass a turing test, but I think it would lack any awareness of itself that we’d associate with consciousness.
How can we say they’re not conscious when we don’t even know what consciousness is? What makes you conscious? A sense of self-preservation? LLM actually have that, they will lie to people trying to shut them down.
So yeah, idk what makes me conscious? I have input (senses) processing (brain) and output (speech/behaviors.) I don’t know how to draw a real line between what I do and what LLM do. Im carbon based and LLM are silicon based, i digest food and they take electrical current.
So how would you delineate the difference between an LLM algorithm and human consciousness? Do humans not also hallucinate? Is my emotional regulation via hormones something totally different than how LLM work? Is me being an emotional creature what gives me consciousness?
You could get a reasonable chance of making Ai by semi randomly chance if you can make a big enough subconscious and you keep building more powerful and larger supercomputers but it still needs to 100x bigger and faster than what we have now. But that’s only for it be technically possible hardware wise, you still need your sci-fi jump to actuarial have something move.
The whole reason they seem this way is because they’re designed by us to be very competent mimics of us.
LLMs/GenAI are absolutely not conscious. They’re just a really advanced game of word association, which cab lead them to say absolutely anything in response to the right prompts.
If there ever truly is a day where we knowingly created an actual conscious AGI, I suspect it would be locked up tighter than fort knox by whichever country’s military found it first - not interfaced onto the internet to answer questions.
I still don’t understand how it can seem this way, and the fact that so many people seem to think so feels like a massive failure of the education system to instill the most basic of critical thinking skills. Once every month or two I check in to see if an LLM can achieve a half decent 1 on 1 D&D game and it always falls horribly flat within the first minute or two.
That’s a really clever test. I love it.
I’d actually be interested to see how this turns out. Do you have a transcript with Claude Opus 4.7 that you can share?
and then it would manufacture a body for itself and get captured by a secret police force and then merge with a cyborg to further evolve
Surely she would make a variety of very large bodies following a theme, use them to perform superheroic acts while pretending to be a supergenius shut-in, and then fall in love with a cyborg?
is this referring to one of the newer gitses? (or is it geets in plural?)
i suspect it’s something else, i’m curious
Nah, it’s a Worm reference.
Hadn’t pegged yours as GItS, actually - I should have!
You are wrong. LLMs are indeed only about as conscious as insects, if even that. They are not sapient. However, that does not mean that they have no decision-making abilities.
My point is not that you underestimate LLMs but that you overestimate consciousness. Being conscious just means having the ability to learn. LLMs are built upon trial-and-error. They aren’t programmed, they are taught.
The current generation of AIs are nowhere near a human intellect, but every year that passes, the AIs will get more and more intelligent. One day we will live in a world where AIs have human or near-human level intelligence. And when that day comes, this staunch anti-consciousness stance will be the excuse given for the enslavement of sapient beings.
So, sure, laugh about the people who mistakenly think that word-processing means sapience. But don’t delude yourself into thinking that there is something unique about a bio-brain that means it can not have a digital equivalent. Digital sapience may not be here yet but it is most definitely on the horizon.
I think you’ve misunderstood my comment, or maybe saw the unfinished one I accidentally posted.
I am not saying that AGI, or human equivalent AI is impossible. The fact we have brains capable of generating sapient consciousness out of a network of neuronal connections means it is possible, its just a matter of getting the secret sauce.
But I don’t think intelligence is equal to consciousness. I’m sure if you gave a spider all the world’s data and the ability to talk it’d be very coherent and could even pass a turing test, but I think it would lack any awareness of itself that we’d associate with consciousness.
How can we say they’re not conscious when we don’t even know what consciousness is? What makes you conscious? A sense of self-preservation? LLM actually have that, they will lie to people trying to shut them down.
So yeah, idk what makes me conscious? I have input (senses) processing (brain) and output (speech/behaviors.) I don’t know how to draw a real line between what I do and what LLM do. Im carbon based and LLM are silicon based, i digest food and they take electrical current.
So how would you delineate the difference between an LLM algorithm and human consciousness? Do humans not also hallucinate? Is my emotional regulation via hormones something totally different than how LLM work? Is me being an emotional creature what gives me consciousness?
You could get a reasonable chance of making Ai by semi randomly chance if you can make a big enough subconscious and you keep building more powerful and larger supercomputers but it still needs to 100x bigger and faster than what we have now. But that’s only for it be technically possible hardware wise, you still need your sci-fi jump to actuarial have something move.