Jeff Hawkins: How brain science will change computing

Поділитися
Вставка
  • Опубліковано 22 тра 2007
  • www.ted.com Treo creator Jeff Hawkins urges us to take a new look at the brain -- to see it not as a fast processor, but as a memory system that stores and plays back experiences to help us predict, intelligently, what will happen next.
    TEDTalks is a daily video podcast of the best talks and performances from the TED Conference, where the world's leading thinkers and doers are invited to give the talk of their lives in 18 minutes. TED stands for Technology, Entertainment, and Design, and TEDTalks cover these topics as well as science, business, politics and the arts. Watch the Top 10 TEDTalks on TED.com, at
    www.ted.com/index.php/talks/top10
  • Наука та технологія

КОМЕНТАРІ • 300

  • @studyviews4572
    @studyviews4572 2 роки тому +5

    This man motivated me to study computer science and neuroscience many years ago. Great to hear his talk again

  • @diy-bunny
    @diy-bunny 4 роки тому +15

    Jeff has nailed it. After 11 years (in 2018), he published it. He is preparing a new book too(Maybe a major update on his old book "On Intelligence").

  • @carloschau9310
    @carloschau9310 6 років тому +32

    This guy is way ahead of his time

    • @amok.4874
      @amok.4874 2 роки тому +1

      No, we’re just behind.

  • @arthur78
    @arthur78 12 років тому +8

    This instantly shot up to one of my top favorite TED talks. Love discovering older gems like this.

  • @MrFrankBullitt
    @MrFrankBullitt 16 років тому +3

    Sharp Guy.
    It's great that he had the courage to follow his interests.

  • @brianjanson3498
    @brianjanson3498 10 років тому +26

    "Some people have fallen into the pit of metaphysical dualism, some exceptionally smart people too, but we can reject all that." LOL.

  • @KaneyoshiSouji
    @KaneyoshiSouji 10 років тому +6

    A lot of inspirational things to think about. Thank you so much.

  • @flochfitness
    @flochfitness 17 днів тому +1

    It’s amazing that I’ve come across this brilliant man many years after this talk after having the same realization about myself: there is no self. “I” is just an amalgam of sensory inputs stored in an organic hard drive in the front of my skull.
    To take Dr. Hawkins theory one step further, our brain isn’t only a predictive machine, it is a mixed reality headset. Our brain has the ability to purposely choose the inputs to take in the keep the predictive model accurate. To keep the game inside our heads going. That game is to create a self, and the world outside the self. This is just a theory, but I am finding plenty of cohesion based on this thought.
    Truly fascinating stuff b

  • @chaz-e
    @chaz-e 10 років тому +7

    Redzo79 Actually this guy is expressing himself freely, unbounded by communication rules, presentation theories and body language concepts.
    Though it's not looking professional, but that's what makes his 'talk' special. :) People are enjoying it and are able to appreciate a novel and a complex idea like "Brain Sc.". What else shall be expected?

  • @Matthitizidu
    @Matthitizidu 13 років тому +1

    Such an awesome point and so obvious, never crossed my mind.

  • @luzic
    @luzic 15 років тому +1

    He's so passionate!

  • @scaleop4
    @scaleop4 15 років тому

    this is excellent.
    thanks for that upload.!!

  • @Jotto999
    @Jotto999 13 років тому +1

    Excellent! Let cognition be understood! It will change our society hugely.

  • @itskittyme
    @itskittyme Місяць тому

    When I saw this video 15 years ago, it gave me a solid foundation on how to think about artificial intelligence.
    And with the invention of LLMs nowadays, I'm always reminded back to this video where he basically predicted this invention of LLMs.

  • @ninjakannon
    @ninjakannon 15 років тому

    I have done exactly that, thanks. :)
    His work agrees with and extends what I'd been hypothesising and trying to do myself, but is far better. I hope to understand how his methods work exactly.
    I haven't seen or read anything recent about his work. He said he hoped to solve the cat-dog problem, do you know how is he doing with this?

  • @stratogustav
    @stratogustav 16 років тому +3

    "If you don't build it, you don't understand it", that was genius.

  • @JamesBlitz00
    @JamesBlitz00 16 років тому

    Would you say it would kind of be a more advanced version of the "predictive text messeging," or somthing else entirly?

  • @benjis007
    @benjis007 17 років тому

    excellent again. thanks.

  • @murderface74
    @murderface74 15 років тому

    Thank you for that.

  • @MohammadAbdulkhaliq
    @MohammadAbdulkhaliq 10 років тому +1

    The future prediction idea makes much more sense than the behavioral concept and frankly the best argument I ever heard against the turing test. But still does not explain consciousness and how it relates to a machine that truly understands or just mechanically always gets right predictions ?

  • @crudhousefull
    @crudhousefull 12 років тому

    Predicting isn't just what it does, though it's a part of it. It also compares current visual/other sensory perceptions (sets of it) to previous sets to see anomalies and then compares it to laws (theories) to see whether they can explain the anomaly. It might also look at all the unexplained data and see if it makes another. Haven't thought about it much, but I'm sure it does many other things. Also, telepathy happens regularly (how many times do you know just before someone calls you?)

  • @yamaha893
    @yamaha893 15 років тому +1

    i had this thought in the mcdonalds drivethrough a while back, so its nice to see a real brainiac go to work on it.
    it makes pretty good intuitive sense if you think about it; the ability to perceive, record, recall and compare patterns encountered by the senses. if you had a perfect memory (perfect recall with the required space to store it all) could one say you had a perfect intelligence (with respect to the experiences you've had) . linguistic difficulties aside, i think it would be so.

  • @YourFirst100k
    @YourFirst100k 6 років тому

    Great edit!

  • @pebre79
    @pebre79 16 років тому

    This video inspired me!!

  • @lordmetroid
    @lordmetroid 17 років тому

    That is what he points out... It is not an algorithm it is pattern storage and recognition. In Neural networks simulation a good result of ingenuity in the pattern recognition can be obtained by introducing noise to the network. Perhaps that background noise is the what makes pattern recignition possible to come up with new ideas and patterns on it's own.

  • @albedoshader
    @albedoshader 15 років тому

    The radioactive waste from fusion processes still has to be dumped somewhere or be further processed.
    Another interesting search term in this field is "Tokamak".

  • @mikecane
    @mikecane 5 років тому

    240p? Isn't there a higher resolution version?

  • @glennmungra5476
    @glennmungra5476 8 років тому

    It's amazing to see how ideas seem to emerge on a global scale around the same time in different cultural contexts.

    • @mwamussa
      @mwamussa 5 років тому

      It's called Universal Consciousness. Invisible, but present. I tap into it everyday by way of precognition. If only you materialists knew about "the other side"...I will leave it at that.

  • @Dirtfire
    @Dirtfire 17 років тому

    You could also check out Numenta's website, which is the company Jeff Hawkins founded to develop his theory.
    Some very interesting stuff they're working on there.

  • @RenePauw
    @RenePauw 13 років тому +2

    Very interesting talk! Although I must say I tend to see the brain more like an antenna. NDE's are very well documented over the past few decades and whether we like it or not, memories can be stored outside the brain...

  • @XxINCHAINSxX
    @XxINCHAINSxX 15 років тому

    Other forms of intelligence are, (Imho) Musical, Mathematical, Linguistic, Literary, Athletic, Artistic, and many more. There are, of course thousands of derivitives of those types of intelligence. Like philosophical...etc...He is very right about what he says, but he focuses on a very small spectrum of what our brains do. Its sad that TED only allows 20 minutes, though it is a blessing as well. Definitely awesome video.

  • @Savaril
    @Savaril 14 років тому

    What a thought provoking argument.

  • @cursorcurs6597
    @cursorcurs6597 11 років тому

    Hard to watch at night, as the intro piece is amazingly loud compared to speeches.
    Hey great work by the way!

  • @therealgeeza
    @therealgeeza 16 років тому

    conciousness is just the centre of narrative gravity for the atcivities of your brain that are needed to effectively intuite the minds of others.

  • @NeuroRevo
    @NeuroRevo 10 років тому

    Actually, most of what you're thinking of is our ability for abstract thought, which allows us to take lessons we learned in one area and apply them to another. We learned how to throw a baseball and so have a good idea how to toss a grapefruit to a friend. While it may seem like it's a new circumstance that requires relearning, it's really just that our brain is able to apply pattern matching at a kind of meta-level.

  • @trivea
    @trivea 12 років тому

    @JonmanXmusic True. I guess he went about it wrong, but I think he means to say that a metaphysical approach from the outset is probably bad because it could lead us to put up false logical barriers (ie. brain can't understand itself so no use trying).

  • @cottreau
    @cottreau 14 років тому

    He is certainly on to something. I am struggling to apply the theory to how I form my current sentence actually. I'm not predicting my own sentence am I?
    I agree, when listening to others or hearing a song or walking along, it's quite similar and I'm predicting my environment based on what I've seen before.
    But what about an artist painting? Me writing or typing?
    I'm curious to see where he goes.

  • @boblolz
    @boblolz 17 років тому

    Awesome.

  • @zeebenzine
    @zeebenzine 12 років тому

    @souravbindia However I also would say that this 'robotic intelligence' that you quote is also definitely a large part of things that we consider as intelligent. But it definitely does not deal with creativity. Actually, I'm interested in what he has to say about the issue of creativity.

  • @usertogo
    @usertogo 15 років тому

    Lookup "Processing in Memory" that's the model that can take the computer closer to the brain. It can already be implemented : massively parallel computing can be programmed in FPGAs. Eventually we will create specialized FPGA logic blocks that will allow to map even large programs by allowing to quickly switch between contexts in memory that is evenly distributed among the programmable logic structures!

  • @JonathanBowen7000
    @JonathanBowen7000 12 років тому +1

    @trivea That, I agree with. I think that attempting to understand the brain is a noble endeavor. We'll be able to actually answer some questions about what makes up a human when we understand the functioning of neural circuitry. But until then, I think we should hold our horses on the implications! Science says nothing definitive about anything outside of it's current capabilities. Jeff Hawkins certainly has a lot of influence, and I feel like he's presenting his own philosophy with his science.

  • @FugalQuease
    @FugalQuease 15 років тому

    cool.now i know where to send my cv!

  • @dowling1981
    @dowling1981 15 років тому

    Where does ESP come into this?

  • @Greenjah81
    @Greenjah81 15 років тому +1

    interesting, love the POV about that our brain is wrinkled because it doesn't fit otherwise 8) Aw well as little question, does our visual memory and dreams are constantly connected with each other, i.e. when we see something in a dream state that we never seen before but know it's exists, where is that coming from? Our memory fantasies or from our visual memory "glitches"?

  • @momentary_
    @momentary_ 11 років тому

    I think the missing ingredient is motivation or desire. The motivation or desire can be anything, but it's prediction coupled with motivation or desire that leads to intelligence. If this is the case, then what we program an AI to want or value will greatly influence it's intelligence.

  • @elpresidio
    @elpresidio 14 років тому

    Read his book "On Intelligence" which goes deeper into prediction using spacial/temporal feedforward/feedback run through 5 senses fed against invariant representations on a heirarchial structure. In addition on each of the HTM levels he goes into the 6 layers and columns of the specific level composed of billions of neurons. L1 mostly of dentrites and how L5 connects to thalamus and the function it plays. He only has 22 minutes to give a preview on this talk. You can also Google "Numenta".

  • @Dirtfire
    @Dirtfire 15 років тому

    Following up on what I said earlier, intelligence actually does increase quite a bit throughout childhood due to a process called myelination, which begins from about age 3 until adulthood.

  • @DSBrekus
    @DSBrekus 13 років тому

    @FamousPhilosopher I agree, its quite refreshing.

  • @lordmetroid
    @lordmetroid 17 років тому

    Good definition of intelligence!

  • @ektrules
    @ektrules 11 років тому

    Intelligence is to perform actions that maximize outcomes. Or atleast, that's the easiest way to measure intelligence.
    One could say reacting to unmatched patterns, unknown situations, and new circumstances, successfully, requires making associations, noticing that the pattern is similar to a previously seen pattern, and reacting accordingly.
    Thinking and imagining can be thought of as continually predicting things. Learning is storing patterns. Creating is a manifestation of things imagined.

  • @Pooua
    @Pooua 13 років тому

    It's interesting that he says that intelligence is the trait of predicting what is about to happen, and how we will begin with such things as intelligent cars. It reminds me that most creatures' predictive abilities are best-suited for linear tasks. Most creatures have very poor exponential predictive abilities. E.g., acceleration is difficult for many creatures to predict, even if it is steady. Some creatures, such as cats, are wired to intercept, even if they should be avoiding, such as cars.

  • @sectionmaker
    @sectionmaker 15 років тому

    bravo..

  • @HierisBestand
    @HierisBestand 11 років тому

    Mind summarizing it for me?

  • @nosson77
    @nosson77 17 років тому

    Although I wont adept to answer the second question I will try to explain the question with an analogy. There is one level of sight, which is bound up with the physical eye. But the sight itself is non physical. However, there is another level of sight that is not bound up with the physical eye (for example when you picture something in your mind with your eyes closed).

  • @ooLevityoo
    @ooLevityoo 13 років тому +3

    Haha, absolutely love his out-of-nowhere stab at dualism :)

  • @mikeyo1234
    @mikeyo1234 13 років тому

    @tanthonight "For example if It wasnt for are desires we would not do anything at all" - My engine management system in my car doesn't have 'desires' but it DOES do something.
    What is the difference between an emotional desire, and a goal oriented programmed device? None in regards to getting things done.
    An engine management system 'feels' the car is cornering left, so it 'decides' to stiffen left hand suspension.

  • @anderwan
    @anderwan 11 років тому +1

    I love the example "they're going to learn, the types of cars the leave their blinkers on for half a minute probably aren't going to turn."

  • @Dirtfire
    @Dirtfire 15 років тому

    I think intelligence is all about prediction and patter-recognition, or rather the CAPACITY to learn invariant patterns in noisy spatial and temporal data.
    The ability to hold several things in your head simultaneously is also the same thing, I think.

  • @boorens18
    @boorens18 14 років тому

    another fundamental peice of brain theory will be explaining how we are capable of metaphor. We communicate extremely often with them and they are key to the brain forming complex theories (you use metaphors to teach)
    Metaphors might also offer a way of grouping memories that are relevant to the current experience by priority, or the most contributive to forming accurate predictions.

  • @elpresidio
    @elpresidio 14 років тому

    @Wescas05. No it's Jeff Hawkins.

  • @mlcrazi
    @mlcrazi 15 років тому

    Hawkins' theory on intelligence seems quite similar to Hecht-Nielsen's Confabulation Theory. I took his graduate neurocomputing class and we programmed systems that have similar memory/comparison/prediction to what Hawkins is saying.

  • @lordmetroid
    @lordmetroid 17 років тому

    You asking me... I am just speculating. As you point out emotions has some part of such decision. The brain could very possibly do pattern matching of emotions as well I suppose.
    What Jeff points out here is that the pattern storage, recognizion and matching is what makes us different from other animals and hence has to be a big part of intelligence.

  • @vif3182
    @vif3182 2 роки тому

    '07 @ 240p
    damn this is almost nostalgic

  • @ForrestDix
    @ForrestDix 11 років тому +6

    16:17 "Oh shit" I haven't laughed so hard in months.

  • @Hallibutbouy
    @Hallibutbouy 15 років тому

    The world is a relational place, the being that is to solve problems, which are a consequence of such relations between things and the subjects relation to these relations, must have a way of relating itself to the things it id dealing with.
    Consider a human born with a paralyzed body (thus without the sensation of touch or the capacity for movement) and without the capacity for emotion but who does possess sight and hearing. What kind of mind or problem solving capacity could we expect?

  • @albedoshader
    @albedoshader 15 років тому

    I know the difference between fusion and fission, but today fusion reactors produce radioactive Tritium, for example. It depends on the shielding material, if it withstands neutron bombarding well or not. And the search for optimal shielding materials is still going on.
    At least Tritium has a low half-life of 12 years, so it radiates not as long as most fission byproducts.

  • @RickeyBowers
    @RickeyBowers 17 років тому

    We should not forget that something having not occured can be just as important as something that did. To say the brain is just a pattern matching storage machine is an over simplification and a way to avoid answering the important questions. How does the reptile brain interface with the storage machine?

  • @PattyBoy31
    @PattyBoy31 15 років тому

    A neat book I read that might be of interest is written by George Christos its called "The Creative Mind: Memory and Dreams" or something....

  • @nosson77
    @nosson77 17 років тому

    You can have an entity with out it being physical. For example a business is an entity it can be brought and sold yet it is not a physical object. So does the business exist?
    I'm not sure if you are against the idea of the existence of the mind in principle or you believe it could be that it does exist but for certain reasons you believe that not to be the case.
    Please clarify.

  • @loveallhumanity2024
    @loveallhumanity2024 11 років тому +10

    Having a lot of information does not always make one more intelligent.

    • @egor.okhterov
      @egor.okhterov 2 роки тому

      Wat?

    • @Acemanveryspecial
      @Acemanveryspecial 2 роки тому

      @@egor.okhterov If you can't process a mountain of information in a time frame that allows you take advantage of its benefits, the information is fundamentally useless to you.

  • @souravbindia
    @souravbindia 12 років тому

    Constant predictions based on matching patterns\ sequences.. that's storing and processing of information.. and recognizing patterns.. that's robotic intelligence.
    Human intelligence is to react on unmatched patterns, unknown situations, totally new circumstances.
    Ability to think, imagine, create and learn during unpredictable events is human intelligence.

  • @jettisoncargo
    @jettisoncargo 15 років тому

    you should watch his other videos. don't forget this is just a 20 min talk he had. there is a 1 hr + speech on youtube where he addresses your questions.

  • @Danelleya
    @Danelleya 15 років тому +1

    omg this guy is great, I could totally imagine how a synthetic brain could be built using these fundamental principles he laid out and would mimic a real mammalian brain

  • @HierisBestand
    @HierisBestand 11 років тому

    Ok thanks!

  • @twobob__
    @twobob__ Рік тому

    2023 and this is so relevant today

  • @kurtilein3
    @kurtilein3 16 років тому

    the problem with the nobel price is that it cannot be given post-mortem. the person still needs to be alive. and there are always much more people that deserve the nobel price than nobel prices to give out. so, sometimes people get their price 30 years later, and/or shortly before they die.
    his talk just makes sense, everything he says about intelligence is true.

  • @alf20104
    @alf20104 14 років тому

    16:24 my favorite part lol

  • @carlossanchez-pg7ij
    @carlossanchez-pg7ij 2 години тому

    I loved❤

  • @Ko252
    @Ko252 13 років тому

    @TheNeurall No, we cant expand with no limitations. One serious limitations is that any choise we make, is based on all the other choises we made before, which led as into the situastion we are in right now, as well gave as the motives, experience etc. that we have. This goes all the way back. So, there is nothing like free will.

  • @dreamdimensions
    @dreamdimensions 16 років тому

    it's all of lifes means of transmitting information

  • @limbworks
    @limbworks 9 років тому

    Just watch this on Netflix and whats eating at me is that he said motor function is controlled in the frontal lobe. Wa.. What! I could have sworn it was further back in the primary motor cortex, brodmann area 4. What motor functions are controlled in the frontal cortex specifically? This is just new to me.

    • @Six_slotted
      @Six_slotted 9 років тому

      chightower26 well imagine if you consciously planned a complex sequence of motor actions that would have to come from the neocortex, whatever parts of the brain actually directly control motor functions would have to translate and action that self created input.

  • @jonesgerard
    @jonesgerard 13 років тому

    I can't escape the suspicion that even with all the new information they still don't have a clue. But that makes it more interesting.

  • @Hallibutbouy
    @Hallibutbouy 15 років тому

    It might be granted that they would develop language but they could have no way of relating themselves to other people or other objects.
    What would give them their purpose for process and output?

  • @nosson77
    @nosson77 17 років тому

    Here is my understanding of the mind. The mind can be divided in to levels, stretching from the brain and going towards the soul and each level has sublevels and each level or sublevel could be simple or elaborate. The lowest level of the mind is logic, going step by step. The level above that is the ability to design, this is the creative ability. The level above this is the level of perception, to see the whole picture. There are levels higher then this but I'm a bit hazy on their nature.

  • @nlamorte90
    @nlamorte90 10 років тому

    dat beemer commercial tho

  • @Radjehuty
    @Radjehuty 15 років тому

    That is what most IQ tests use as a foundation when measuring intelligence, however pattern recognition is a vastly oversimplified and narrowed view of the capabilities of the human mind. You could consider "multiple intelligences" or "crystallized vs. fluid intelligences" as another perspective.
    Just because the most IQ tests test mainly pattern recognition, does not mean that Intelligence is defined as such.

  • @DavidBall67
    @DavidBall67 16 років тому

    John Ball's Patomic Theory actually predicts how the brain works, but it also has been used in application and may allow computers to translate better than translators.

  • @XboxSeriesX_Player
    @XboxSeriesX_Player 2 роки тому +5

    kto od Quebo?

  • @Hallibutbouy
    @Hallibutbouy 15 років тому

    firstly because it may come in many different forms that appear not to need the same abilities, secondly there are problems arising in who judging cases of intelligence. Obviously there is the subjective aspect of what is considered intelligent by that judge, but more obscurely whether the being who is being judged as intelligent needs to be conscious of what they are doing. When biologists remark that a 'virus is intelligent' what exactly do they mean? That it is good at problem solving?,

  • @lailatulqurrotaayunin8754
    @lailatulqurrotaayunin8754 3 роки тому

    Damn my dude is ahead of his time

  • @George4943
    @George4943 15 років тому

    Having been a Mensa Proctor, I am familiar with what is tested when IQ is determined.
    IQ (if not 'intelligence') is, in fact, (mostly) pattern recognition.

    • @egor.okhterov
      @egor.okhterov 2 роки тому

      All intelligence is pattern recognition

  • @JonathanBowen7000
    @JonathanBowen7000 12 років тому

    "Some people say 'Brains can't understand brains'"
    I agree with them.
    It's not just "very zen-like." Assuming the mind and our understanding is limited to functions of the brain, unless there's some new neuroscience I don't know about the brain can be understood as a logical system. We can create our own systems of understanding understanding, but we can't understand understanding on a level greater than understanding itself.

  • @nosson77
    @nosson77 17 років тому

    Occam's razor is used on two competing theories. Lets have the theories in full and then we will see which one the razor cuts off. And by the way only you can loose because I said "what if". So, are you ready?

  • @DavidBall67
    @DavidBall67 16 років тому

    Something else. It works on a theory of how the brain processes the written grammar, not the computer designed filter approach. Computers are much faster than brain synapse. There is a description at thinkingsolutionsdotcomdotau

  • @souravbindia
    @souravbindia 12 років тому +1

    Human brain is the most advance and most complex system in the universe explored till date. The brain has evolved through millions of years, its built inside the womb in weeks, everything really is programmed within genome.
    Philosophically, the reason to create universe is to create a intelligent creature like human!
    Its like creating a robot who can create another robot!!
    He is looking for an answer which many other persons have been researching on for thousand of years.

  • @Johnsson25
    @Johnsson25 5 років тому

    Ролик назван по-русски, но без перевода? Вот спасибо.

    • @tSpline
      @tSpline 5 років тому

      надо включить субтитры

  • @Dirtfire
    @Dirtfire 16 років тому

    And this new thing he's working on will be bigger than that, even.
    It's not the Turing Machine that we've all used since the 1930s; This is something fundamentally different, and I believe Hawkins will have just as prominent a place in history as Alan Turing.

  • @Hallibutbouy
    @Hallibutbouy 15 років тому

    Very interesting, I find him fresh, well motivated and targeting the right areas.
    However, it is a pitty that all nueroscientists and modern philosophers of mind that I have come accross are ignorant of the body and its relationship with the world.
    A brain makes no mind without a body and the bodies existence is entirely determined by its interaction with the world.
    Good Luck!

  • @Xx_Venom_Fox_xX
    @Xx_Venom_Fox_xX 15 років тому

    What constitutes intelligence is highly controversial with many believing that non- cognitive functions, e.g. impulses and emotions, play a role in "intelligent behavior.". Predictions based on a memory system are frequently flawed as are memories. It is not that "we are constantly making predictions" so much as appraisals of our perceived relationships to others and the environment. You cannot totally disengage personal interest from the pursuit of knowledge and problem-solving
    activities.

  • @nosson77
    @nosson77 17 років тому

    MightyTiny I think to say that the mind is the brain is unfounded. I think the only reason why people say it is, is because they believe in materialism but if you can see a distinction between hardware and software why would you constrain your self to say the brain doesn't have that capacity. Especially if you realize that the more something is bound to the physical the more limited it is, compare the mechanical calculator to a modern computer