Jeff Hawkins: The Thousand Brains Theory of Intelligence | Lex Fridman Podcast
Вставка
- Опубліковано 2 чер 2024
- Jeff Hawkins is a neuroscientist and cofounder of Numenta, a neuroscience research company. Please support this podcast by checking out our sponsors:
- Codecademy: codecademy.com and use code LEX to get 15% off
- BiOptimizers: www.magbreakthrough.com/lex to get 10% off
- ExpressVPN: expressvpn.com/lexpod and use code LexPod to get 3 months free
- Eight Sleep: www.eightsleep.com/lex and use code LEX to get special savings
- Blinkist: blinkist.com/lex and use code LEX to get 25% off premium
EPISODE LINKS:
A Thousand Brain (book): amzn.to/3AmxJt7
Numenta's Twitter: / numenta
Numenta's Website: numenta.com/
PODCAST INFO:
Podcast website: lexfridman.com/podcast
Apple Podcasts: apple.co/2lwqZIr
Spotify: spoti.fi/2nEwCF8
RSS: lexfridman.com/feed/podcast/
Full episodes playlist: • Lex Fridman Podcast
Clips playlist: • Lex Fridman Podcast Clips
OUTLINE:
0:00 - Introduction
3:04 - Collective intelligence
9:46 - The origin of intelligence in the human brain
22:59 - How intelligent life evolved on Earth
33:58 - Why humans are special in the universe
37:16 - Neurons
41:30 - A Thousand Brains theory of intelligence
50:10 - How to build superintelligent AI
1:08:10 - Sam Harris and existential risk of AI
1:20:12 - Neuralink
1:27:02 - Will AI prevent the self-destruction of human civilization?
1:32:34 - Communicating human knowledge to alien civilizations
1:42:50 - Devil's advocate
1:47:45 - Human nature
1:56:07 - Hardware for AI
2:02:46 - Advice for young people
SOCIAL:
- Twitter: / lexfridman
- LinkedIn: / lexfridman
- Facebook: / lexfridman
- Instagram: / lexfridman
- Medium: / lexfridman
- Reddit: / lexfridman
- Support on Patreon: / lexfridman - Наука та технологія
Here are the timestamps. Please check out our sponsors to support this podcast.
0:00 - Introduction & sponsor mentions:
- Codecademy: codecademy.com and use code LEX to get 15% off
- BiOptimizers: www.magbreakthrough.com/lex to get 10% off
- ExpressVPN: expressvpn.com/lexpod and use code LexPod to get 3 months free
- Eight Sleep: www.eightsleep.com/lex and use code LEX to get special savings
- Blinkist: blinkist.com/lex and use code LEX to get 25% off premium
3:04 - Collective intelligence
9:46 - The origin of intelligence in the human brain
22:59 - How intelligent life evolved on Earth
33:58 - Why humans are special in the universe
37:16 - Neurons
41:30 - A Thousand Brains theory of intelligence
50:10 - How to build superintelligent AI
1:08:10 - Sam Harris and existential risk of AI
1:20:12 - Neuralink
1:27:02 - Will AI prevent the self-destruction of human civilization?
1:32:34 - Communicating human knowledge to alien civilizations
1:42:50 - Devil's advocate
1:47:45 - Human nature
1:56:07 - Hardware for AI
2:02:46 - Advice for young people
It seems to me that my dog looks at me for help all the time, so does a dog realize that we are more intelligent than they are?
Last conversation was Epic, really looking forward to this one. Thanks Lex
You should save everything from the short speeches you start with someties, and after a while let a AI cut it, in regards to what it think will be profound/inspirational. Great as usual, read his first and heard the first episode - seems like his research really are breaking new ground, will read it. Love!
Lex are you really tryna give computers rights before animals? Computers aren’t even conscious(yet). I think it’s beautiful that you care about the potential for computers to suffer but maybe we should focus on the systems that we already know are suffering today(animals). Looking forward to your vegan guests!
Hello Lex, are you alright? Yes I do understand your mental health struggles, dealing with depression as a teenager, but it got better as I learned more about the world.
I enjoy lex's interview style so much. He asks a question or makes a point then let's his guest expound without constantly interrupting his guest.
Just a reminder to the young viewer, this is the man behind the only viable 'PDA' (Personal Digital Assistant) to date, the Palm Pilot. Ergonomically excellent due to it's compact size and simple stylus driven UX and ingenious Graffiti writing 'language'. Killed by a dev(i)ce that even today, is ergonomically inferior and unlike the Palm devices, does not inspire creative thinking. Jeff understood how using both hands to poke a keyboard ties up the whole brain, whilst using a pencil / stylus frees up the creative side. Too few product designers today study human factors. My Galaxy Note 9 is the closest to UX perfection in a device. Anyway, a distraction from the subject of this video, but wanted to ensure people know about Palm. :)
Highly relevant recognition - helps put his significance to the conversation in perspective. Always suspected if anyones intuition could be justified on the subject his would be in the top percentile. Like you said it took us, what, another 30 years to match his achievement experientially? [At one point I had 3 Palm Pilots, still have my LifeDrive]
@@Kobe29261
Got a fully working Tungsten T3 and a Treo 650 in the draw right next to this keyboard :D
Obviously have plenty of modern devices, but these two gems will never leave my collection.
@@Blowfeld20k Respect! A man may be so far ahead of his time, his age finds little to no use of his inventions, like all the old artifacts they keep digging up for which nobody has an explanation. Most of those men were probably thought crazy. Da Vinci's journals is a perfect example; no appetite for half what the man dreamt up. We won't have to rewrite the history but anyone who cares will find this here man one of the first to intuit [way ahead of dead Old Steve] that a keyboard is about the highest coefficient of friction you can place between a human and a computer. Why it still beats me that people didn't swarm Samsungs styli-enabled devices as soon as they were released! If our age is lucky Elon might erase the chasm between intention and execution to neuronal conception! What a time to be alive!
@@marcus8710 did more than your smart phone? can you be more specific? these were before my time, i just looked up a review, and honestly it looks a good deal less usable and user friendly than a smart phone.
Interesting.
Jeff's book On Intelligence was the only book I ever found in that era that actually answered some questions about what intelligence meant and how it worked. I never found another person who read it so I had no idea if it had any impact in academia or whatever, so it's good to see him here and being taken seriously by someone at least.
Yes, this area is treated as somewhat niche for some reason. Of all the areas currently treated as potentially revolutionary, this is one of the most underrated.
it's probably because of much of what he is saying are hypothesis that aren't verified through experiments, but are just his hypothesis and framework of human brain. They're all very interesting, but for it to be taken seriously it has to verified by experimental results.
“On Intelligence” was way ahead of its time. Kahnemann’s “Thinking, Fast and Slow” is always touted as a great book on the brain, but it’s just super boring in comparison.
Jeff was always ridiculed by neuroscientist a decade ago because he had no formal training in biology. Glad to see him gaining the respect he deserves, and Lex inviting him on for the second time.
Sometimes I feel like I’ve learned so much more from UA-cam videos like this than I ever could have in school.
I didn’t understand how I learned until I went to college and took german history. I suddenly realized I learned by creating a story, for a story for everything. The danger is in getting the story wrong.
The second most important realization about learning was taking care to ask the question in the right way, often revealing the answer, or at least revealing the nature and peeking into some fundamental truth of what you are trying to understand. Both realizations have held up and served me well.
Now realize you can share that story through words and pictures and make stacks of cash 💸
I had never clicked on anything as fast as I did on this one. Thank you Lex! A Thousand Brains is truly amazing.
I was just thinking the same thing! Been a while since I've heard from Jeff Hawkins. He was hugely inspirational to me studying computer science
Ha! I did the same. Thanks, Lex.
I watched Jeff’s Ted talk 15 years ago in High School, made me incredibly interested in AI, computers, the brain, and even talked to a couple people at his company Numenta. When he talked about legacy and what not, I found that part interesting. I’ll probably never meet the guy, but his research and talks have had a profound impact of my life goals.
Thinking of phantom limbs makes me realize how powerful these predictive models can be. You could be missing a limb and still feel it there even though there is no input. Or how somebody with ptsd will perceive threats when there are none because their prediction model is oversensitive.
Phantom limb syndrome and associated pathologies are a great clues to what's going on in the brain. Its like research you wish you could do but no ethics committee will ever approve (and rightly so lol) .
Wow very true
100% correct. concisely put, thank you.
What a fantastic discussion. I really liked that you disagreed with each other, makes it more interesting as a viewer as it makes you both explain why you think something at a deeper level. And disagreement and discussion can only lead to learning something :)
Its because Lex didn't read his book and asked him to explain the whole thing lol.
The opening is “classic” Lex Fridman: beautiful, epic, seamlessly merging philosophy, futuristic science and the every day. Amongst the best, a new breed of intellectual, Humboldtian. My great respect.
🤓
I would LOVE to see Jeff Hawkins talk to Lisa Feldman Barrett.
Yep - this is like her theory of construction
Very pleasant, informative, thought provoking conversation with Jeff Hawkins, whom I had not heard of until today. Thanks, Lex, your podcasts are, indeed, a great service to humankind, and who knows, maybe even non-humankind in some future scenario.
I love how passionate this guy is about the topic
This is fascinating! I will re-listen to this many times!
Lex, thank you for all of the knowledgeable guests!! I learn so much intently listening to this show! Very grateful!
Lex never misses
This discussion I found maybe the best insight into neuroscience and AI - thankyou so much for this world knowledge artifact. Great thinkers!
When I was younger when I learned something it felt like there was always a space on the right shelf just waiting to connect with predetermined relevant filing system, more like revealed knowledge.
48:30 Craziest thing for me is when someone makes a mistake speaking or I read something wrong and my brain calculates at a mile a second a bizarre justification for the statement, before they correct themselves or I re-read and it's actually something totally normal. The fact my brain is able to jump through so many hoops in milliseconds is astounding.
Thanks Lex. You really have created something special with this podcast. Another great one!
This guy is incredibly interesting to listen to. His passion is captivating!
cheers to your channel Lex...hard to find a better interviewer out here.
Fascinating conversation!
It was so great listening to two people who are truly passionate about learning and discovering the very complex issues of life!
I like this guy so much, Appreciated Lex, FINALLY someone talking good about AI FINALLY.
LMAO ikr
I like the return to AI as well, but this didnt feel like a conversation. Jeff seemed a bit rehearsed in his approach, almost like he answered direct passages from his books and sometimes these answers didnt even relate to the questions
Good job on the Codeacademy sponsorship. I learned Java through that website and it was one of the best learning experiences I've ever had :D
Great show. I really love discussions on the brain and the way it functions.
I just finished reading this book it’s brilliant indeed! Thanks Lex.
Just went and watched episode 25 to prepare for this one.
Not gonna lie, I’m excited to see this simple, logical theory application blow past all of those NLP, AI vision theories in the first few years. Segmenting experience is not experience and does not lead to learning. Sensory/motor multimodal experiences can lead to deep, meaningful learning, knowledge and existing.
The knowledge and perspective in this conversation are something to truly be appreciated.
this has been one of my favorite episode, maybe only behind the one with ian hutchinson and nuclear physics.
Awesome! I follow Jeff’s work on his channel love the approach. Thanks for another great video
I can’t believe he said “We’re not just individuals ‘we live in a society’ and so on”
Groundbreaking
It’s a good argument to guard some knowledge. Example: You don’t have to be a clever hacker anymore; you just need to purchase the program and aim.
referencing pegasus? en.wikipedia.org/wiki/Pegasus_(spyware)
@@dankoni -interesting and absolutely maniacal in its application and use. My wife has a diary, but at no point would I ever dare open it and violate her privacy. Some Dark Tech, and evil in the wrong hands with bad intentions.
With GPT-3 you simply have to ask. The barrier of entrance is creativity and curiosity I think.
Cheat codes lmao.. They hack what is allowed..
Boy, that was a beautiful podcast!
Loved "Hedgehog in the fog" reference ))
Ezik v tumane
Lex love your podcasts. Follow each n every one of them, so interesting n informative. Love from the Fiji Islands 🇫🇯
OMG bula! So nice to see a fellow Fijian into this as well!
Jeff is great at explaining the complexity of the human brain. Now i just want to finish reading his book!
ur site is pretty cool, keep posting esp about brain...
@@yt-sh thanks!
This sparked a question. Could the urgency of mortality be a story we tell ourselves? to fit what's happening in our life with the world?
Isn't it a spectrum like most things. When life expectancy was much lower, was our urgency exponentially greater? Does doubling our life expectancy half our urgency to get stuff done?
Yes. There is no consciousness. It’s just pattern recognition. When we look at an ant, we think we are intelligent and they are dumb. That thought is so stemming from a more complex brain. There is no you.
Lex's little "mhmmm"s" n "yeah's" are cool
Another really great podcast Lex! - I will get Jeff's book and send him a note . .
I am so pumped for this! Great work Lex 👍
The meaning of life is recognizing it in the moment. I experience it everyday, in so many ways.
I love this idea that the outcome of our physical laws when the universe was created a consciousness to gaze back at the splendor of it all.
Love your podcasts lex thanks so much and keep it yoo😇🌎
Great interview. Bought the book because of it. Can't wait to read it
How did you guys not even touch on the hard problem of consciousness? Lol, great podcast either way!
This was a fascinating discussion to me. I think I'll have to rewatch it though fully digest....
Profoundly grateful for your opening words!
looking at learning as creating a map of something feels very natural. Especially with music and the feeling of getting lost and not knowing a way home.
Thanks so much Lex, another great episode...
Wow another spectacular podcast, these podcast are becoming my daily lunch routine. Your ability to ask the right questions is unbelievable and as a fan I’d love for you to make a video with Illia Polosukhin. Illia is a computer developer who worked for many company’s including google and now is the co founder of NEAR protocol which is an amazing project in the crypto space. He has so many insights into the world of computer developers and I’d love to hear his perspective. Thanks for the educational podcast, I’m grateful for all of the different types of guests you get on with such different backgrounds.
Nice explanation of the modelling and motive systems.
How about the model of self? Is that an old function?
Where does the formation of habits fit in? The meta model of the ideal self can overcome some instincts.
Very interesting conversation. Thank you Jeff. Many thanks Lex.
Great podcast. I like Jeff more every time I see him, and he seems to have a real handle on the intelligence thing.
Woohoo! I've been following Hawkins and Numenta closely for a decade.
I think Jeff's comment at the end was spot on: Lex, you are doing a great things with these interviews (paraphrasing)
I love Jeff Hawkins analogy of the modelling method very interesting indeed. ,Great interview, thank you gentleman !
Thanks Lex
Jeff is one of my favorite thinkers in the AI space and I really enjoyed your interview.
Fascinating. Jeff’s way of thinking reminded my neurons of JOHN DEWEY’s pragmatic approach to intelligence, learning, and the problems of philosophers (e.g. free will).
1) 15:40 Of course, Dewey is famous as the founder of progressive education, based on “learning by doing.” The 1KBs theory seems to closely track many of the key insights of Dewey’s educational psychology, as developed in the Lab School he founded in Chicago in the 1890s - and later summarized in his 1916 book, Democracy & Education.
2) Dewey lacked Jeff’s understanding of neural networks, but his description of the interplay between the decision-making that emerges from the ongoing competition among multiple mental maps/ideas below the level of conscious awareness, and the small but vitally important redirecting efforts that occur at the conscious level, has always struck me as an obvious solution to the so-called “problem of free will.”
In Human Nature & Conduct (1922), he describes this process in terms of two systems, impulsive and structured (i.e., modeled/mapped). Very similar to Dan Kahnemann’s “Systems 1 & 2” model.
3) Dewey often noted that most problems of philosophy aren’t solved but are gotten over - that is, we see sufficiently far beyond the traditional dualism (freedom/determinism) such that the problem dissolves or is seen as an uninteresting impediment to learning. (e.g., I can’t identify a single novel idea from Sam Harris on free will. Augustine covered that terrain pretty effectively 2k yrs ago. Nietzsche pointed out in 1887 that “thoughts happen” is a more accurate description than the subject-centered “I think.” Etc.).
Great show! Fascinating 🧐
Self-replication vs intelligence. One's risky, the other is not.
This blew my mind a little bit.
I disagree with this separation. I think sufficiently advanced AI could figure out self-replication while solving a different task, as a sub-goal.
Wake up babe , new lex Fridman podcast
Thanks for the work you do.
You the man, Lex!!
Thank you for yet another thought-compelling podcast! ☺️
I know for me, whilst my comment isn’t directly relating to the topic at hand, now I am wondering what are the motives for some humans that want there to be evidence of their existence to other life forms?
Needless, thanks for stimulating my mind. Much appreciated!
Motives include community building where there is an exchange of useful knowledge and capabilities the knowledge provides, especially if we find ties beyond the stars to similarity. Some for story too which is often worth hearing. Humans possess one if passed to right person at right time changes everything so linear time is constant. Use it wise and always. In some cases knowing how to build relationships between what you know and what is universal is a key skill and applicable to all intelligent life forms. Humans dare there to be greater intelligence than themselves but organizational intelligence and cybernetic combination of knowledge of a workforce directed under a common visible goal is formidable to individuals so community building, social, behavioral and cognitive superhuman advances can move forwards. A life form that can exist after a cataclysm, the disruption of our physical world
Interesting talk! Better reminds me that I want to make help people feel better in the world we live in.
Reminds me of your interview with Joscha Bach. He blew my mind with his same explanation of intelligence being the ability to "make models". I understood it right away. He didn't have to explain anymore. I quickly built the model of what it looks like to build models in the mind, and it blew my mind how spot on that definition is.
About to order this "Thousand Brain" book now too. 😆
I'd like to comment on the introductory word, about preserving something about humans, from historical perspective.
In the late III millennium BCE, there has been a ruler of a Mesopotamian city-state of Lagash, called Gudea. He was one of the most powerful figures of his time, bur unlike other rulers, never called himself a king, only a governor. Similarly, instead of documenting his wars like all the others, he documented his manifold building activities. Those documents were written in perfect Sumerian - a language that was at the verge of dying out, perhaps not even spoken at the Lagash any more. The text were directed to the gods, and were not to be seen by mortals.
Of the temples Gudea once built, nothing has remained. But 4 millennia later people from the land unheard of in the Ancient Mesopotamia, mortals with the powers of the gods, have discovered Gudea's texts about those buildings. Those texts allowed for the deciphering of the Sumerian lanuage and with that gave us a much better understanding of the history of the region, and, broadly speaking, of who we are. In a way, Gudea's dream came true.
If we leave something after us, and someone else gets a grip on it, it will almost certainly not be used, or even understood as we intend it to. But if we do - we live.
very helpful. tyvm
There's a Hibiscus flower on a Fiji water bottle? Flower components... stamen? What. My brain compressed the hell out of that...😂
I've watched the first interview two weeks ago and I loved it. I found quite a bit of worthy stuff in that interview so I'm eager to see what more I'll learn here.
Great podcast, just trying to bump the algorithm
Well I was going to sleep. Lol Must listen. 👌 Love ya Lex.
Awesome introduction Lex!
Please more AI episodes. I love it. 1k brains
I really enjoy the way Jeff Hawkins thinks.
btw, you da man Lex, the new Rogan of YT. Keep it up.
2:44 "The extremes of conscious experiences" Wasn't that the motivation of the Cenobites?
Ha ha ha ha! Fuck, good point.
@@johanfritz8209 Lex is actually Pinhead... explains a lot.
1:48:00 Disappointment - Absolutely.
~58:00
"Is it possible to build intelligent system which doesn't care about turning it off ?"
Now I want a debate between Jeff Hawkins and Elizier Yudkowsky, moderated by Lex, specifically about convergent instrumental goals.
Be fascinating to have Jeff Hawkins and Geordie Rose for a round 3 Interview.
Hi Lex, greetings from Boston. Nassim Nicolas Taleb at some point in the future? Great conversation as always, thank you
I understand now why Joseph Rogan says your the best! ;)
Lex is my private indulgence in intellectual conversation
not anymore...
Thanks Lex and Jeff for this great discussion and a lot of information. Made me think that we are here living great lives in harmony with nature integrated in a system because of love but may be made extinct only by selfishness. We may build super intelligent machines that learn love and maximize its capability to bring great experience and purpose to our lives.
The best way to understand the brain and the body and the way it works is to look at it when it goes wrong.
Thank you so much.
Best into thus far. Stance 🙏
Thanks Lex, I just watched "Heghog in the fog" I recently, also now experienced the machine elves.
Lex, first of all thanks for these interviews. I want to suggest taking a fabric steamer to that black backdrop. Would get those wrinkles right out and make that studio even more sharp. 👌
"It seems prediction is fundamental to how our mind operates" - Lex my man, You I think had spoken on a previous podcast on Conciseness. If we are to believe and I do. that signals from my toes take longer to reach brain then the signals from my vision. And I do. Then we are already living in the past in a buffer.
Because we are already working behind real-time data. we must attempt to act as-if we are in order to do things the best we can.
My opinion is when athletes and I'm sure other domains have this. but the concept of 'being in the zone' is equivalent to enough practice that you have offloaded the tasks to something closer to the data. and your intelligence in the task, or as Jeff might say. your Model of the task is so up to date. You can then do it quicker, and better than those who must still create a model as good as your own.
This comment is prob long enough to not get read o// :)
Lex keep up the good pod's casting into my face.
Wonderful, introduction. The most important of human experiences, love, the closest we can be to the divine.
I could not agree more, well said, as you so eloquently phrase, the most vital aspects of our human existence.
Another great show
I must get his book. To get a thousand times smarter 😃🙏 Great interview 👌
- You are really productive .. !!!
Thankyou for another great interview Lex.
During the section you labeled “How to build a superintelligent IA” and then in the following existential risk section, the idea seems to arise (implied) that building these multiple copies of the mini brains will produce a conscious system. I think this is not going to happen, as Mark Solm points out in his book The Hidden Spring A Journey to the Source of Consciousness, the cerebral cortex operates at a subconscious level and that the seat of consciousness is in a structure at the top of the Brain stem. It would seem to me a different type of neural structure will be needed to add consciousness.
Nice, few days ago watched first podcast.
Lex - I would argue that part of what makes us human is love, which is by nature self-sacrificing. Can an AI be self-sacrificing? You are doing great work I thoroughly enjoy your interviews thank you.
I love your work lex! Keep it up! I envy you !... 👌
It's not a surprise that complex systems are actually really complex.