I'm convinced LLMs "evolve" in the way Bob says they do. Team Paul and Steve: 0; Team Bob: 1
@noahghost44763 ай бұрын
Yeah but the question is ultimately about free will, isn't it? And it's about whether the ability of the LLM to learn teaches us anything about human cognition. It doesn't. You can't design a simulation of human cognition and use it as evidence that human cognition is some kind of way. We have not solved the Hard problem of Consciousness. (Chalmers). So we are standing outside the door, unable to even enter discourse about the fundamental nature of human cognition. Llm has tokens and attention layers, iterative processes, all kinds of complicated functions, so that it can recognize patterns and use multiple levels of analysis and use language in a way that is indistinguishable from language that might come from you or me. But just because you see the same Moon doesn't mean you took the same path to the top of the mountain. There's more than one way to skin a cat, and when you do, it's not climbing any mountains. Bob, this is not just an exercise in futility. We are here to stay, so we have to steer the ship toward Heaven instead of hell. Don't be an aloof nihilist. My favorite part of your Buddhism book was when you pointed out that Buddhism is not necessarily a kind of nihilism. If the nature of things is that we are here to stay, it matters a lot what we do. The thicket of views, I think. And that dude that kept asking questions in the Buddha would not answer. It's not correct to just think there is nothing. There's obviously something and we are here to stay. The thing that finds itself existing is a being. The evolution of God is not anthropology. It's the evolution of this disoriented being find itself existing and tries to make sense of things. And tries to make it better.
@Lolleka3 ай бұрын
The intro is too funny. :D Also, loving Paul's mug c:
@FavoriteMovieIsBooks3 ай бұрын
I don’t meditate but when I jog I always focus on my foot falls to distract myself so maybe I do sort of meditate
@boydhooper40803 ай бұрын
Mike Tomasello clearly and concisely explains the evolution of language. He Answers all of the sticking points discussed.
@AnaArOes3 ай бұрын
We don't have to forget that we are conscious in a world that it is always present. So our conscience doesn't can much in empty space with no others, etc.
@Zidana1233 ай бұрын
<a href="#" class="seekto" data-time="1220">20:20</a> What a month of meditative retreat does to a man
@fuadsafarov652 ай бұрын
Dear Bob, why isn't there an audiobook version of your first book called "Three scientists and their gods"? Please, everybody, like this, so that he can see.
@derekwilson79193 ай бұрын
Paul has a nice coffee cup. How does one get such a cup?
@timsebert94913 ай бұрын
I tune in for the ball breaking, not the intellectual curiosity, it's the ball breaking.
@NicholasWilliams-uk9xu3 ай бұрын
Dude Paul is so awesome and smart, this is good stuff, thanks.
@elliotreeves75793 ай бұрын
Please for the love of leopald dont make these shorter
@paulk38303 ай бұрын
Hi Bob I recently learned that when coming back from a meditation retreat, all you have to do is listen to the past few "the seven" podcast from The Washington Post.
@AnaArOes3 ай бұрын
Thank you.
@jimkozubek40263 ай бұрын
Free will exists. Ideas pop into your head, and you have the ability to ignore or strengthen your commitment to any given idea. As such, free will exists to the extent of you cultivate ideas.
@theoshouse82153 ай бұрын
If determinism is true (which is what both Bob and Paul believe and what you believe if you believe in Einsteinian physics) then you /don't/ have the ability to ignore or strengthen commitments to thoughts. You just happen to do one or the other (ignore or strengthen) in accordance with your desires, which are exogenously determined (by your environment, genetics, education, et cetera).
@noahghost44763 ай бұрын
@@theoshouse8215 there's probably something I misunderstanding. But it makes no sense to me that Neo darwinists assume it has to be an unguided process. We experience ourselves as conscious entities, for Christ's sake. What sense does it make to assume there is no intentionality or teleology or whatever it's called. One way or another, we are beings having experiences and guiding processes. One way or another, it's not naive to think there might be something akin to a mind that plays a role in guiding evolution. Everywhere I look, people are making arguments about how the theory of evolution is just falling apart. Like the irreducibility of complexity in the bacterial flagellum. Google that, that's pretty crazy. I think it's naive to jump on the determinism bandwagon. It's just a giant assumption basically that reality is the way it seems. Isn't that the most naive assumption? Isn't that the age old, classic naive assumption that is constantly being made and we never learn? This whole idea of determinism is rooted in us childishly believe in the nature of reality is the way it seems to us. Might as well join the Flat Earth society. I think the hard problem of Consciousness brings us back to the drawing board.
@jimkozubek40263 ай бұрын
@@theoshouse8215 It's hard to believe that consciousness or self-awareness is determined. that's why it feels spooky and disjointed from the world.
@henriknederman3 ай бұрын
Confirmed. Bob think about sex, just like the rest of us.
@AnaArOes3 ай бұрын
The grammatical sintax is given by objects in the world, that not change, it is not located in the brain. The body uses voice, ears, eyes to capt reality, and the vocal apparatus to mention objects and actions. An the neurons work functionally.
@robertgreenwood22583 ай бұрын
Sentences don’t exist anywhere out there in the world.
@Lolleka3 ай бұрын
I mean, it's a noisy world. Lots of the noise is correlated. You call AI training "evolution", but really both are mixed stochastic-deterministic optimization processes. There are many ways to describe the same exact thing. The difference is in the expressive abilities of the physical substrate and the process that drives the optimization. For biological entities it is some form of natural evolution (but not limited to). For AI it is backpropagation and stochastic gradient descent.
@savyse13 ай бұрын
So Skinner was right after all? 🙂
@rightcheer50963 ай бұрын
Good luck canceling Musk.
@geaca32222 ай бұрын
Wow the excuses made around <a href="#" class="seekto" data-time="500">8:20</a>, haven't we learned from the me too movement how abuse of power works? The victims lose their job and career opportunities if they don't comply. It's traumatizing.
It's also male victims, not only girls/women. I was interested in this conversation but the topic of sexual harrassment and how it is handled as some kind of joke and serious excuses made for it, makes this video unwatchable for me. So thanks for removing this opportunity for me to learn about a topic. Apart from the popular meat market por--n bots that pop-up everywhere. Only people who have been victimized by sexual harrassment (not about sex but about power!) know how their voice and detrimental horrifying experience is silenced. It happens in small companies too.
@jamescoll1303 ай бұрын
Unlistenable.
@Kevin-dj3cc3 ай бұрын
First 😃👍
@olewetdog62543 ай бұрын
So you think you're better than us?
@olewetdog62543 ай бұрын
Sorry but having that many children is NOT something to praise. The world is crowded.
@JimStanfield-zo2pz3 ай бұрын
You also have to consider parsimony. If processing over a simple single algorithm gives you a robust approximation of the many more complex algorithms that together give rise to complex behavior than evolution is likely to find that single exact algorithm of the many exact algorithms. The hard nativists aren't t quite dead, but they are definitely ill and not looking too good 😂. There are probably a few other simple algorithms yet to be discovered, but the idea that are genes are encoding many complex algorithms into us from birth up to the level of complex conceptual models of reality is definitely on its death bed. And there is no way of ressurecting it.
@bofbob13 ай бұрын
Can't say I share that impression. At least in linguistics (the field where I work), it doesn't look like the "hard nativists" are at all concerned by any of this. Which is fair enough. From their vantage point, there's really no there there. Which is how you get someone like Chomsky saying that LLMs are a feat of engineering but they add nothing to scientific knowledge. Some people have dismissed that as him just being snide, but from his perspective it's exactly true, a simple statement of fact. Really it's just large theoretical ensembles colliding, each drawing on a different set of conceptual metaphors, but with neither really having an edge over the other in terms of explanatory power. Which is business as usual in linguistics. Truth is we know fuck all about language. ^^ That's what makes it fun. Well, for me at least. 🙂
@JimStanfield-zo2pz3 ай бұрын
@@bofbob1 true, I don't actually fully believe my argument, I believe that the deep learning method are in fact discovering both the priors and the finer details, what we call learning, at the same time.
@JimStanfield-zo2pz3 ай бұрын
@@bofbob1 the only thing I would say is that the rationalists for really no incredibly good reason were overly hard on the empiricists for a long time mostly motivated by Noam Chomsky. That's why I think they deserve some jabbing. Sure there are many priors when it comes to human intelligence and language acquisition, virtually no empiricist ever really argued against that. There has never been anyone that ever argued the blank slate model. Instead of actually seeing the validity of many of the credible arguments that the empiricists were making the rationalist, i.e. Chomsky and Pinker, invented a straw man, tabula rasa, and argued against that. There has just been too much intellectual chicanery from that side. And it's about time empiricism made a comeback, they were always making the more parsimonious argument. Essentially that evolution may be providing some macro scale structure that allows that brain to parse out some fuzzy signal within the noise, but that learning was basically the primary way the brain fills in the detail.
@bird3341783 ай бұрын
The meditation retreat Bob went to did wonders for his mood.
@danielhansen70273 ай бұрын
Leopold
@catherinemaddux2413 ай бұрын
Bob you seem in this conversation the exact of someone who’s been on a meditation retreat. Very combative!
@johns.72973 ай бұрын
EP rules. Thank you Darwin.
@noahghost44763 ай бұрын
Paul, good news! I like you guys again. I'm sorry I called you a woke professor. Youre both wicked hilarious. I wanted to stop watching after the first 10 minutes in case you said something to piss me off, but I'm glad I kept watching; thanks for that discussion of the evolution of language. About saving the good stuff for after the paywall: it's fine to float that idea to plant it in the brains of the audience, but really the highest quality content should be before the paywall, because that's what gets people to subscribe. Please tell Bob (because I know you're reading this and he's not! That aloof m'ffr!) to try putting a different category of stuff after the paywall. It could be anything... for example, you can analyze the same topics through the lens of Buddhism or your work on empathy, or some psychological theory, or through the lens of some conceptual model you guys make up together & let the audience be part of. It could even be a guided meditation. Some kind of experience for the audience. Or an opportunity for the audience to learn a particular thing thoroughly. In overtime you can give people a kind of "take away." Btw, near the end I see Bob saying people should not follow their retributive instinct. That's not what Israel is doing! I don't know how you guys can be so smart about all this other stuff and be so wrong about israel. I see Bob joking, like, he told the policymakers how to conduct a war on terrorism and they didn't listen. But there's an important Insight contained in that joking around, bob, because as awesome as you are you do not understand escalation dominance! Your criticisms of Israel are ridiculous. Bad things need to be done in war. That's true regardless of whether my saying it makes you dismiss me as someone trying to feel tough or hawkish or whatever. The highest priority has to be making sure kidnappers can't control situations by kidnapping. That's why Israel has that Hannibal policy, you have to do whatever it takes to discourage evil. Okay that's all for now good talk.❤
@Other-rc7mw3 ай бұрын
What Paul is describing at <a href="#" class="seekto" data-time="2740">45:40</a> is not free will, but mere will. Sure, we have the capacities that he refers to, but we don’t have the capacity to oversee those capacities, which is what free will (as conventionally understood) would need to consist of. Compatibilism is a solution to a problem that doesn’t exist. It’s perfectly possible to describe our deterministic reality in honest, straightforward language that doesn’t imply things that aren’t really true. Sam Harris (a fool on nearly every topic, but a sage on this one) showed the way in his book “Free Will.” I would suggest that Paul give it a closer look.
@MikeDrewYT3 ай бұрын
Darn! I forgot to send Paul my Ask You Anything questions.
@AnaArOes3 ай бұрын
I mean we are nothing if the world isn't there. To remember something of the world must be present.
@JimStanfield-zo2pz3 ай бұрын
It could be that evolution primarily acted on traits that generally improved the performative aspects of informational processing in the brain which coupled with improved mapping of auditory voaclizations to environmental signals. Yes, after humans developed languange im sure evolutionary pressures had an influence on the propagation of mutations associated with overall lignuistic ability. However, evolution may have been primarily acting on cognitive traits that generally improve cognition. Like improving the metabolic processing of fats and sugars into energy in the brain, yes this would improve language ability, but it would also have improve cognition across a broad range of domains. The fixation of such a mutation may have been driven primarily by its impact on language, but it was not just a language influencing trait. There could also be domain specific mutations that impact only language processing, but it could be that those mutations are harder to find. One reason to believe this is because there was clearly a basic brain archetype that evolution was already acting on before humans came a long. Its akin to microprocessors 50 years ago, all of the basic computational components were there, but the overall performative capacity of those components was not, but the path forward was simple, add more transitors, increase cache size, increase the function of the arithmetic logic unit; this is a far easier way to improve than thinking of entirely novel architectures to improve the system.