I’ve been going through Eagleman’s book “Incognito” this week and boom, robinson interviews him. Great guests you’ve been having on, congrats and much more success to you!
I genuinely wish I didn't like the taste of coffee. I've been off it for several months now and I generally feel much healthier for it. But even hearing you talk about it brings back the craving.
I don’t doubt the guest’s expertise. However, I’m wondering if he’s being a bit too absolutist in his comments. I know at least one person who learned to speak Chinese as an adult and did so with no accent (according to native speakers). Re: memories before the age of three. I have distinct and unpleasant memories of my experiences learning to swim at 18 months old.
“I know at least one person who learned to speak Chinese as an adult and did so with no accent (according to native speakers).” I agree that adults can learn a second language with no accent (although it’s probably pretty rare) and can give an example that anyone is able to check out. _See_ Aran Kim of _Aran TV_ here on this site. She was born and raised in South Korea; Korean is her first language. She first learned English when she was a *junior* in college (in other words, well past the age of 13). She has no accent-in fact, you can see a video of hers “Foreigners Guess Where I'm From Based On My English | Reaction To ‘I'm Korean!’” and _everyone_ thinks she is from the States.
Yes! I'm definitely hoping to speak with him. You must have been watching a lot of recent episodes, because animal minds have been on my own a lot lately...
Robinson I wanted to recommend the book "A Foray into the worlds of animals and humans" by Jakob von Uxkull. You'll definitely value this, it's a deep dive into animal consciousness.
@@robinsonerhardt totally. He's also the guy that coined the term "umwelt" in the early 20th century I believe, to describe the "lived world" of the organism.
44:51 “My question is what if we had invented Sora-this text-to-video generation-before Isaac Newton came along, before we had F=ma, D=1/2gt², all this way of looking at physics that we humans do? What if, instead, we had done it the way that squirrels look at physics, where they've just somehow got intelligence about physics but they're not, you know, making it into equations.” Really, there’s something embarrassingly inane about this speculation. That’s just the distinction in psychology between _rule-governed_ (e.g., formulas) and _contingency-governed_ behavior. People as well as squirrels _always_ operated according to contingency-governed behavior-they could catch a ball arcing high in the sky long before they could extract the formula for a parabola that would describe the ball’s motion. Sora’s text-to-video generation is, similarly contingency-governed-Sora’s not calculating the physics of waves crashing on the beach, it’s arranging pixels that emulate patterns of those showing waves crashing on a beach. If “we had done it the way that squirrels look at physics,” nothing would have changed because, until Newton came up with his laws of motion, we _were_ doing it the way squirrels looked at physics. It says something about how siloed academia when someone in a field, neuroscience, that is closely related to psychology knows apparently nothing about a distinction that has been talked about 80 years.
audio still bad :( i try to listen to the pod, even at low levels, and when the ads hit, they are SCREAMING, they are SO loud, compared to the levels of your recorded audio
If people like conspiracy theories, they have done something bad to their brains during their lives. If being correct had any survival value, people would not follow conspiracy theories. If being part of the group is more important, then demeaning yourself by believing stupid crap might have survival value if it lets you into the group. I question whether that survival would be worth the trouble.
I have to point out that the argument about avoiding meat/animal products has more to do with the relatively cruel practice of factory farming than an animal’s theoretical ability to suffer.
Hes exactly right about LLm's they can only answer questions corectly that have already been answered - they idea that they show emergent behaviour - is a compleatly false illusion created by manipulation of the metrics being used to measure them see - Sanmi Koyejo paper on the subject
They are in a sense the ultimate plagiarists, taking everything they have scan, putting it in a tumble dryer, then spitting it out the other end and saying hey look what I've created!
I disagree with the conclusion he made about LLM models, that they are only answering from the data that has been fed to them, hence there is no creativity or emergence of new stuff and that's why they are not conscious. Because the same argument could be made about us humans, all the so called creative ideas we come with could be an amalgamation of the knowledge we collected since our birth be it visual, audio or any other experience. I SERIOUSLY DOUBT that we are coming up with a unique thought of our own. I believe that whatever ideas we come up with is actually an amalgamation of the inputs we had since our birth. They appears to be unique cause the amalgamation includes ideas made by extremely unrelated inputs which other people might not have experienced in their life. What do you all think.
@@RajatAgarwal-h5l i think your asking were do ideas come from - but the point about llm's is they cant answer questions that havent already been answered and the answer is recorded on the internet . - but humans can and do answer questions that have never been answered . They also ask new questions that have never been asked.
I disagree with the conclusion he made about LLM models, that they are only answering from the data that has been fed to them, hence there is no creativity or emergence of new stuff and that's why they are not conscious. Because the same argument could be made about us humans, all the so called creative ideas we come with could be an amalgamation of the knowledge we collected since our birth be it visual, audio or any other experience. I SERIOUSLY DOUBT that we are coming up with a unique thought of our own. I believe that whatever ideas we come up with is actually an amalgamation of the inputs we had since our birth. They appears to be unique cause the amalgamation includes ideas made by extremely unrelated inputs which other people might not have experienced in their life. What do you all think.
I want to start by saying I can absolutely see where you are coming from. I don't want this to degenerate into ad hominem attacks or one upsmanship which unfortunately is all too common in yt comment sections. That said I would imagine the audience for this channel would be a little more mature and capable of engaging in a productive conversation on what is a deeply fascinating and perplexing question. I do hold a different belief than you but by no means am I sure I am right. I think definition of terms is quite important. I'm going to use the phrase 'novel ideas' to describe what I believe a sentient human mind is capable of , and which differentiates it from something like a generative large language model chat bot. It's just a plain fact that the sum total of Novel creations, let's say science fiction for example, Can be treated like branches of a tree where each branch origin can be regarded as a conceptually novel idea created by a human. There was a first person who thought of time travel either into the past or the future. We can debate the semantics of whether it was ever truly original or derivative but nonetheless I don't believe a large language model could spontaneously come up with that concept before it had encountered it as part of its own training. Again, it's just my opinion but I genuinely believe we are still missing something critical when it comes to ideas such as self awareness and creativity when trying to engineer intelligent automata. I'm not one of those people who believes sentience will emerge by throwing ever greater capacity and performance into the existing paradigms we use today. We are missing something qualitative in nature and until we discover that piece of the jigsaw puzzle I don't think we have to worry about giving artificial minds their own set of human rights. This is a topic that deeply fascinates me and in fact I have written a science fiction novel over the last 4 years which explores this very subject amongst a litany of others. With all that said there was a previous interview on this very channel with a neuroscientist who I think presents a compelling case that we are essentially organic machine minds, and I tend to agree with his position. Like I said I just still think we are missing something very important, And most likely more than just one thing as well. I seriously doubt we will ever get to witness an engineered sentient machine mind in our own lifetimes and perhaps not ever. My novel is a little more optimistic :)
“I disagree with the conclusion he made about LLM models, that they are only answering from the data that has been fed to them, hence there is no creativity or emergence of new stuff…” It’s not the inputs that are the issue-it’s the process that determines the outputs. In the case of something like AI coming up with, say, new drug molecules, the AI is just coming up with different plausible combinations at a rate many times higher than people can. It’s not really “creative” (although these AIs are coming up with novel combinations)-it’s just what David Eagleton calls “level one scientific discovery.” As for LLMs, they’re designed to come up with the statistically next most probable word, more or less (to allow their answers to have a bit more variety and appear more human-like). That, in a way, is almost the _opposite_ of creativity, although, again, their specific outputs might be novel in the sense of never having been produced before. Whatever human beings are doing, they’re _not_ coming up with the next most probable word. The other thing is that, as Eagleton points out, it’s hard to imagine an LLM coming up with something novel along the lines of Einstein’s theory of relativity (if such a thing had never been produced). Why? Well, again, they’re basically designed to emulate verbal statements that have been made before. But also they have _no_ (and I mean *no*) insight-aside from their architecture, their responses show they have none. (I’ve gotten answers from LLMs that are not just wrong-they _couldn’t_ be right with just a moment of reflection.) So, even if an LLM _could_ come up with a statement about riding on a beam of light, as Einstein did, it wouldn’t know what to do with it. (Not that LLMs really “know” anything, anyway.) People are, of course, different. No one can say _how_ Einstein came up with his thought experiment of riding on a beam of light-even if Einstein could, it’s not clear that his verbal report would accurately reflect how _he_ came up with it-but once he did, he could work through the implications. I doubt he’d be going through previous _verbal_ formulations he had heard, which is what, in essence, an LLM would do. I’d surmise it was some kind of unconscious process drawing on his entire environmental history, something LLMs don’t have (except, again, for the extensive verbal inputs) and can’t do (again, they don’t have insight). That’s my 2¢, anyway.
2 EDGES SWORD will say, our Son of Man carrying thee! Likewise ye know? Thy 2 EDGES SWORD can't pass beyond this point. But Ye alone! My shared "i" Am Forefathers Gratitude and Honor!
Gnashing of teeth come here in front and remind! Lord flesh and blood are not thy enemies! But these principalities who deceiveth exalted themselves above sitteth in high places unseen nor seen in front of thee! Ye have sent thy messenger called Time before HIS VISITATION!
