back to article Samsung is planning to reverse-engineer the human brain on to a chip

Industrial mega-conglomerate Samsung is working to "copy and paste" the structure of the human brain onto computer chips. The Korean megacorp has linked with Harvard University to replicate the make-up of the brain in a chip format in the hope that doing so will allow future chips to access the brain's "low power, facile …

  1. msobkow Silver badge

    Well, we hope that if we mimick well enough, a miracle will occur and our hardware will actually be intelligent. We have no reason to expect this to be so, as apes and other human-similar brains aren't "intelligent" and "aware", but we're out of ideas. We're chasing the old theory that "any sufficiently complex system is implicitly intelligent."

    1. Anonymous Coward
      Anonymous Coward

      Strongly disagree that non-human brains aren't intelligent or aware.

      1. This post has been deleted by its author

      2. karlkarl Silver badge

        Agreed but I also disagree with the OP that human brains are intelligent and aware in the first place.

        1) As a human myself I am making a self deprecating joke.

        2) Awareness is very much a man-made construct for complex decision making and inference. How would we know if we are truly aware? Yes, you can jump on the spot or touch your nose to "prove" it. But a machine could just as easily do that too.

    2. Imhotep Silver badge

      They are also assuming that we know enough about how the human brain is built and how it works to replicate it.

      That sounds somewhat delusional right there.

      Wasn't it just two years ago that they discovered a new human organ?

    3. Imhotep Silver badge

      The article reminds me somewhat of those people who say that human and animal studies are no longer needed because now we can do it on computers.

      Yes, we could if we knew everything absolutely everything about the human body and how it works to program it for a computer.

      But if we knew that, we wouldn't need the studies anyway.

      1. msobkow Silver badge

        We also wouldn't need those filthy humanses any more.... *muttering*

  2. G R Goslin

    Let's hope

    Let's hope that it's not going to be anyone from the UK Government

    1. My other car WAS an IAV Stryker Silver badge

      Re: Let's hope

      Or any of the Group of Seven (G7) + Russia + China + Oz and plenty more -- any national gov't past a certain point of bureaucracy.

    2. DJV Silver badge

      Re: Let's hope

      I could easy model the brain capabilities of several of our "esteemed" politicians - just pass me that lump of cold cat sick...

      1. JetSetJim Silver badge
        Coat

        Re: Let's hope

        tsk - massively overengineered. Or are you going all of them at once?

    3. Death Boffin
      Paris Hilton

      Re: Let's hope

      Whose brain did you use?

      Abby, Abby someone.

      Abby who?

      Abby Normal.

      1. Donchik
        Mushroom

        Re: Let's hope

        Genuine People Personalities GPP anyone?

        Ghastly...

  3. devin3782

    Fast forward some years - What began as a conflict over the transfer conciousness from flesh to machine, escalated into a war which has decimated over a million worlds, the Apple and the Samsung have all but exhausted the resources of a galaxy in their struggle for domination. Both sides now crippled beyond repair, the remainance of their arms battle on ravaged planets, their hatred fueled by 4000 years of stock market manipulation.

    1. VonDutch

      Wow, that takes me back. Guess I'll be hunting that down to play tonight.

      Coincidentally the game was 24 years old yesterday!

      1. devin3782

        Yeah I love to play it every now and again, its spiritual successor supreme commander never quite captivated me in the same way.

        1. VonDutch

          Forged alliance yes. SupCom2, really no. There's a community group that are still modding and improving FA (Forged Alliance Forever).

  4. gecho

    They should start with a cat, when it completely ignores everything you try to teach it they can claim success.

    1. msobkow Silver badge

      But then they'd need a "puke in shoes" bot for the AI to run...

      1. The Oncoming Scorn Silver badge
        Coat

        That's a interesting boot-strap problem.

  5. Peter D

    Alternatively

    They could carefully rear babies in a nurturing environment, send them to the best schools and universities and offer the most promising ones a job at Samsung when they're in their early 20s.

    1. Twanky Silver badge

      Re: Alternatively

      (Homer mode)

      20 years?

      I want it now!

      (/Homer mode)

    2. Dan 55 Silver badge

      Re: Alternatively

      It'd never work, they'd realise they could do so much better than Samsung.

  6. Mage Silver badge
    Boffin

    Really?

    How will they do this as we don't really know how a brain works?

    Loads of creatures are intelligent, some by inference of human biased tests appear to be self-aware. Many animals including crows, bees and primates have vocabulary, but so far only humans seem to have language. See Chomsky. Some have claimed parrots or chimps are communicating using language, but independent experts have not verified any language skills, only vocabulary. Language lets you construct new ideas which can be tested or simply entertain.

    Also the coarse structure is one thing, but how that actually works is different. Also a computer neural network is a sort of dataflow HW & SW distributed implementation and not much like biological neurons which are only part of a brain. A fruit fly has about 100,000 neurons. A human has about 90 billion neurons. But neurons are only one part of brain structure. You might fit the structure of a fruit fly on a chip using billions of transistors. Except we don't have enough detail on even how that works to "copy and paste" it on a chip.

    This is simply research and has very little to do with how a biological brain works.

    "Although neurons are often described of as "fundamental units" of the brain, they perform internal computations. Neurons integrate input within dendrites, and this complexity is lost in models that assume neurons to be a fundamental unit. Dendritic branches can be modeled as spatial compartments, whose activity is related due to passive membrane properties, but may also be different depending on input from synapses. Compartmental modelling of dendrites is especially helpful for understanding the behavior of neurons that are too small to record with electrodes, as is the case for [the fruit fly]"

    This is not really what they claim it is.

    1. vtcodger Silver badge

      Re: Really?

      Upvoted because you clarified some things I wondered about.

      I agree that it's seemingly not really what is claimed. But it sounds like an interesting project none the less. Probably things will be learned. How do they plan to "train" the thing? It's not like humans pop out of the womb equipped to discuss philosophy or quantum mechanics. We need a few years of preparation.

      My one quibble is your first paragraph. Animals may not "have language" whatever the heck that means. But many are quite effective communicators. Dogs, for example, manage to convey their desires and opinions far better than many humans. And most understand the limited components of human speech that are of interest to them. That portion involving food, walks, etc. They do not seem to give a damn about climate change, Brexit, or whether the 2020 presidential election was stolen from Donald Trump. Makes them pretty bright in my book.

      1. Mage Silver badge

        Re: quite effective communicators

        Yes, Bees, chimps, Corvids, dolphins etc all communicate. Even if a creature has a vocabulary or spoken words or chimp sign "language" that just means it can communicate. Language is a step above vocabulary or communication. I didn't explain it well. See Randall Monroe's Explainer book and Noam Chomsky.

    2. Twanky Silver badge

      Re: Really?

      Some have claimed parrots or chimps are communicating using language, but independent experts have not verified any language skills, only vocabulary.

      Wossname, wossname, wossname!

  7. b0llchit Silver badge
    Pint

    Copy... paste... infringement... Profit!

    ...Samsung is working to "copy and paste" the structure of the human brain...

    But if you copy a human brain, then you are infringing copyright. Not only the copyright of the brain's creator, but also all the original brain's thoughts and produce, concrete and potential. The (owner of the) copy is guilty of instantaneous copyright infringement at an closely to infinite scale. The lawyer domain will have a field day retrieving near infinite times the maximum fine qualifying to the entire wealth of the universe. And it is true, the value of this single copy of a brain is more valuable than any universe.

    1. msobkow Silver badge

      Re: Copy... paste... infringement... Profit!

      Oh? You created your own brain and have copyright or patents on that?

      Proof please. This should be VERY amusing...

    2. The Oncoming Scorn Silver badge
      Childcatcher

      Re: Copy... paste... infringement... Profit!

      Professor MacLaine, you are being served with a Cease And Desist Notice, your IP address has been logged as frequently downloading brain wave patterns, infringing on the IP held by the copyright owners including but to limited to Astronaut, Fighter\Test Pilot, Neurosurgeon & storing them in on some form of as yet unidentified portable storage medium, you keep close to you, sorry would you mind keeping your boy out of your office while I am discussing the legalities of this take down notice with you.

      1. Mage Silver badge

        Re: Copy... paste... infringement... Profit!

        Joe 90!

    3. Anonymous Coward
      Anonymous Coward

      Re: Copy... paste... infringement... Profit!

      That cuts both ways. For example, everything you learned from Mickey Mouse is owned by Disney, and if you are not paying royalties on that, your persona could be subject to a takedown notice, with large fines and criminal penalties if you don't comply.

  8. katrinab Silver badge
    Meh

    A couple of problems:

    If you just copy the hardware, you get a dead brain, or, if you are lucky, a 0 day-old baby.

    We don't know what the difference between the brains of for example Donald Trump or Dominic Raab, and a human of normal intelligence.

    Do we even know if all the important functions actually take place within the brain?

    1. msobkow Silver badge

      Well, if the goal is to emulate Trump's "intelligence", they can just plug some wires into a brick and claim that it "works" by sitting there doing nothing. All it needs is an APT-3 module to complain about how everyone is "against" the brick and "persecuting" it...

      1. adam 40 Silver badge

        Another Brick in the Wall

        Trump's wall!

        1. msobkow Silver badge

          Re: Another Brick in the Wall

          *ROTFLMAO*

    2. Anonymous Coward
      Anonymous Coward

      I'm somewhat Autistic (HFA) and so keep generally up to date with what goes on in neurology because I have something of a personal interest in it, and I find it somewhat interesting.

      Most of our knowledge about what parts of the brain do come from various people who have suffered penetrating physical trauma destroying part of the brain eg Phineas Gage (https://www.smithsonianmag.com/history/phineas-gage-neurosciences-most-famous-patient-11390067/)

      The best summation of our current understanding is that we are at the stage of a scientist circa 1700 with no understanding of electricity trying to reverse engineer a modern 10nm CPU with a candle and a knife.

      Scientists think that all of the important functions take place within the brain although they can't even agree on fundamental basics of what the brain does and how it does it. For instance; the brain might store memories etc by interconnecting between lots of neurons (a mix of a deduplication system and a relational database) and we don't really have concepts for a lot of things going on. For all of the lofty aims of this sort of research we are still at the stage of trying to understand fundamental basics at the moment.

      We are not anywhere near the stage that we can actually achieve something like this unless i've missed about a millennia of research being done in the last year or so much the same comments apply as to Facebooks attempts to create a neural interface; We don't know 1% of the information required to make it work but the abysmal and spectacular failure will certainly provide some useful information we don't have at the moment before the person funding it drops the funding.

      If it was possible to copy and paste a brain into a computer and run it then I would suggest that the reproduced brain would essentially be imprisoned in a state of sensory deprivation and is unlikely to cope at all well with these conditions; I certainly wouldn't volunteer to be a test subject.

      1. Mage Silver badge
        Alert

        It will be claimed to be a success!

        An IBM computer wins at Jeopardy. They claim this is advanced AI (it's nothing of the sort). They then sell a Medical AI system for millions with same branding. Advanced AI to spot cancer or something? But it's unrelated software and data to the system that won Jeopardy and also doesn't work.

        At the end of this something will be marketed. However it won't at all be a copy of a brain. Or a Brian.

        1. Anonymous Coward
          Anonymous Coward

          Re: It will be claimed to be a success!

          About the Medical AI system failing to make money - that doesn't mean it couldn't have been useful, just that it didn't make money for medical industry, which is a different thing.

          Ocassionaly the NY Times runs a health article focusing on some patient with unusual debilatating symptoms who spends years getting incorrect diagnoses or being told it is a psychological problem before finally getting lucky with a doctor who actually figures out what it was because that doctor had a wide base of obscure medical knowledge. And usually some other doctor writing in accompanying comments section also recalls learning about that in medical school.

          But cases like those are not a money maker, because they don't come in bulk, and because they are obscure there is no fear of malpractice for not solving it.

          On the other hand a new cancer medication that has been "proven" (by the drug maker) to extend life for three weeks can make a fortune, because it can be used in a huge number of cases.

          I think it is quite likely that Deep Blue could transfer that Jeopardy ability to become a useful medical tool for holding obscure knowledge that doctors could use to iteratively input symptoms and get recommended tests.

          To actually make that product viable, IBM would have spend money on very smart doctors to work with the AI engineers during development while working through a lot obscure cases (which would likely also include a lot of cases which did have a psychological basis). Then, with a portfolio of successful cases, they could market it to health providers, who would use that as a means of advertising - maybe.

      2. Imhotep Silver badge

        Have you ever read The Origin Of Consciensness In The Breakdown Of The Bicameral Mind by Julian Jaynes? A brilliant and fascinating book with a title that seems designed to frighten people off.

        In the lead up to his main thesis, he gives the best description I've ever run across of how we think and of how earlier peoples located some functions we ascribe to the brain to other parts of the body.

        I ask the question because of your first two paragraphs.

        It is one of the few books I go back and reread every year or two, and I am always amazed at how often I'll read a new paper or even an old history that helps to support his arguments.

        1. Anonymous Coward
          Anonymous Coward

          I'm interested in biology/neurology; that book appears to have been written by a shrink without significant knowledge of either history or science. Yes, we are utterly certain that the ancients physical understanding of bodies was massively defective and our assumptions are made by more logical processes.

          The ancients made reasonable assumptions based upon their knowledge and the fact that they didn't do dissections etc. For instance, there was a long standing assumption that the heart was responsible for emotion, which is based upon the common feeling on a breakdown of a romantic relationship of almost physical pain in the centre of the chest where the heart is.

          We can now identify the general areas of the brain responsible for certain things via MRI scans etc and nobody actually advocates the theory that the heart is responsible for emotion, although even after hundreds of years the heart symbol still has cultural significance on valentines day cards etc.

          If you are interested in learning more then for the historical understanding of the body and medicine the best work to start with would be The Story of Medicine, by Kenneth Walker FRCS which is a great read aimed at a layman and telling a story of medicine from the earliest of history through to when the book was written.

          For a more or less modern understanding of what parts of the brain do what I would suggest Neuropsychology: The Neural Bases of Mental Function as a pretty good starting point as being deigned to introduce medical students to the subject.

      3. LybsterRoy Bronze badge

        One of the books I've been reading recently is "Reading in the Brain" and that supports you comment concerning learning about the brain from the injured ones - basically seeing what's lost or recovered.

      4. JetSetJim Silver badge

        If it was possible to copy and paste a brain into a computer and run it then I would suggest that the reproduced brain would essentially be imprisoned in a state of sensory deprivation and is unlikely to cope at all well with these conditions; I certainly wouldn't volunteer to be a test subject.

        Agreed in part - if it were posible to copy/paste the state of a developed brain, then the bits that are expecting visual/audio/touch/taste/digestive/... inputs would certainly have a somewhat disconcerting experience, and this may have an impact on that silicon-brain. Similarly, severing mechanical ties to movement may also have a bad effect - I wonder if there might be a similar amount of pain experienced to those that have limbs severed with their "phantom limb" syndrome. Would there be a "phantom body syndrome"?

        So, how long would it take this silicon construct to readjust itself to its new confines - could it sit as a higher layer protocol on a computer and interface to a web-cam for eyes? Or would an optic nerve interface need to be developed (see Night Walk). Similarly for audio reception.

        On the whole, it would probably be easier to get input into the "brain" than to extract output as our commuication is inherently controlled by a physical process. Would we need to also incorporate some form of nervous system that interfaces to loudspeakers?

        On the whole, rather challenging before you get down to the ethics of it. Technically you may well be creating a new person with the paste operation - the personalities will instantly start diverging.

        Does the copy get to vote? Is removing a power source from it murder?....

      5. Kibble 2
        Stop

        "If it was possible to copy and paste a brain into a computer and run it then I would suggest that the reproduced brain would essentially be imprisoned in a state of sensory deprivation and is unlikely to cope at all well with these conditions"

        The research on this would take a great amount of time and many steps in my estimation. Also there is the probability that the contents of a complete brain would be copied and pasted without training in accessing available networks is slim. I would guess that creating a virtual world prior to powering up the personality would be a necessary step.

        If you would like to consider another philosophical point, would turning the program off be considered murder?

        1. JetSetJim Silver badge

          > I would guess that creating a virtual world prior to powering up the personality would be a necessary step.

          You;ll need to have to figure out how to pipe sensory inputs into the brain, first, and somehow figure out how the brain can provide output into that world, also. You won't be able to just 'jack the brain into an Unreal Engine and expect it to be a seamless interface

    3. Someone Else Silver badge

      We don't know what the difference between the brains of for example Donald Trump or Dominic Raab, and a human of normal intelligence.

      You're making a dangerous assumption, there, katrinab, that either of the two you mentioned actually have one in the first place.

      1. katrinab Silver badge
        Meh

        Well Trump can recognise a KFC burger and work out the correct way to lift it to his mouth.

        Computers really struggle with that sort of task.

        1. JetSetJim Silver badge

          The dragonfly has only 16 neurons and is one of the most effective hunters on the planet.

  9. elsergiovolador Silver badge

    Abyss

    When you gaze into the abyss, the abyss also gazes into you.

    Imagine having a sandbox environment for your ideas or just having a spare brain for others to pick, while you can focus on perfecting a reverse spliff.

  10. Scott Broukell

    I think, therefore I am, waiting for an update to complete, please do not turn me off whilst the update is in progress.

  11. John Savard Silver badge

    I'm Enthusiastic

    Well, if this could be turned into a prosthetic third hemisphere, to be implanted (with some kind of battery) in human brains... so that after our flesh and blood bodies died, we could go on living forever, it would be a really good thing.

    1. Mage Silver badge
      Alien

      Re: I'm Enthusiastic

      You've read too much Iain M. Banks. It's fiction, not a blueprint. You also listening, Mr. Musk?

      1. Killing Time

        Re: I'm Enthusiastic

        IMB was certainly not the first float those ideas. Frederik Pohl (Man Plus) and William Gibson (Johnny Mnemonic) just to name a few....

    2. Anonymous Coward
      Anonymous Coward

      Re: I'm Enthusiastic

      Well, if this could be turned into a prosthetic third hemisphere, to be implanted (with some kind of battery) in human brains... so that after our flesh and blood bodies died, we could go on living forever, it would be a really good thing.

      Would it?

      The entire of human society revolves around the old eventually dying off and passing their assets to the next generation. Would you care to guess the results of what would happen if all of the elderly opted for being uploaded to a server rack in the living room and continuing to live?

      Especially as many of them would be able to keep working too; as if you could work remotely then you could continue to do so when dead if you were uploaded to a computer.

    3. Kibble 2
      Coat

      Re: I'm Enthusiastic

      https://www.amazon.com/Altered-Carbon-Richard-K-Morgan-audiobook/dp/B0009I6MC8/ref=sr_1_1?crid=1UTO9IFTVC4O7&dchild=1&keywords=altered+carbon&qid=1633021878&s=books&sprefix=altered+%2Caps%2C231&sr=1-1

  12. Dr_N Silver badge
    Terminator

    Call-Me-Kenneth says,

    "Get ready, Fleshy Ones..."

  13. steelpillow Silver badge
    Boffin

    Aww cute - trying to run before we can walk, woopsie!

    There is a project going in the UK to map the brain of the humble bee, with much the same purpose in mind. Smart insects, bees. Among the smartest, I believe. The next step will be to implement it on a chip, as a proven evolved neural network, so as to find out its secrets over the crap we simulate today, and use that to create next-gen networks.

    A lot more plausible than mapping my hairy fizzog and setting my clones on your business model.

  14. Ken Hagan Gold badge

    "The brain is made up of a large number of neurons, and their wiring map is responsible for the brain's functions. Thus the knowledge of the map is the key to reverse engineering the brain."

    That sounds like a hypothesis we need to test, rather than a predicate for future work. But hey, if someone else is paying...

  15. David 45

    Who's brain?

    Let's hope they don't try and reverse Trump's brain - they would have difficulty in finding it in the first place!

  16. Twanky Silver badge
    Coffee/keyboard

    What's that...

    ...splashed on the keyboard? --->

    What do you mean 'the last guy who tried Ctrl/Alt/Delete'?

    Ah. I for one welcome our new...

    1. Jimmy2Cows Silver badge
      Terminator

      Re: What's that...

      Sounds like an outcome of BOFH 2.0 - The Upgradening

  17. 45RPM Silver badge

    If the human brain was simple enough for us to understand then we’d be so simple we couldn’t.

    I’m beginning to suspect that might not be true anymore. But, having said that, whilst I don’t fear the day that computers are more intelligent than us I’m bloody terrified of the day that they’re only as intelligent as us - although, in fairness, I am conflating intelligence with rationality.

    1. Denarius Silver badge

      its time

      to read Roger Penroses two books, The Emperors New Mind and Shadows of the Mind. Another futile attempt because of oversimplified analogies. WARNING: Both books assume intelligence of reader. Some knowledge of set theory and basic understanding of Godel helps.

      1. steelpillow Silver badge
        Boffin

        Re: its time

        The trouble with theories like these is that there are too many untested foundational assumptions, with most of the conclusions waxing lyrical on the consequences of those untested assumptions. For example, the microtubule model described in Shadows is regarded as disproved, or at least lacking in the slightest credible evidence, by mainstream biology.

        On the other hand, there are good points which do need to be taken on board, such as the non-algorithmic nature of brain function. Von Neumann machines do algorithms, neural networks generally don't. We can look at an algorithm and go "oh, gawd, I suppose I'll have to process that myself", but we are never very good at it.

        And there are things missing. For example might chaos theory and massively interconnected parallelism between them cover for the non-algorithmic aspect (c.f. state machines), no magic quantum weirdness required?

  18. adam 40 Silver badge

    Emergent properties

    I'm thinking (!) if you shovel enough compute power and memory together, then something will emerge from it.

    Whether it will be intelligence we'll have to wait and see. It might take a few iterations of different wiring and rewiring between components, subsystems and inputs and outputs to get it to respond to its environment, but once it starts responding then you can start training it.

    1. JetSetJim Silver badge

      Re: Emergent properties

      You could have a silicon brain the size of a planet, but if the code reads:

      10 PRINT "I AM AWESOME"

      20 GOTO 10

      you will not in any way achieve "intelligence".

      I doubt anyone is anywhere even close to replicating a single neuron in code, let alone how umpteen billion of them interconnect, take inputs, produce outputs, grow, adapt, change

      1. Jimmy2Cows Silver badge
        Boffin

        Re: Emergent properties

        Did you reverse engineer Trump's operating system?

  19. Doctor Syntax Silver badge

    We already have brains. We may not understand how they do it but we have a good understanding of the results it can produce. A much better effort would be to look at what the human brain finds difficult and make products that help with those tasks. But that sounds remarkably like what we've been doing all these years with so much existing software.

  20. Derek Kingscote
    Boffin

    What you need...

    What you need are a bunch of updated Transputers [remember them?]

    If you're a youngster, wikipedia is your friend.

    Old hands will probably share their experiences

  21. RLWatkins

    Ha ha ha ha ha ha....

    ... ha ha, ha ha, haha!

    I'm so very sorry, but this is one of the stupidest and most blatant "Hey, let's all pay attention to ME!" press releases I've seen this year.

    Not only do we not have the technology to do this, we don't even have the technology to design and manufacture any technology which might even remotely approach the necessary capability. Not even close.

    How on earth can anyone take this seriously?

  22. big_D Silver badge

    Thou shalt not

    As Butler said, in Dune, "Thou shalt not make a machine in the likeness of a human mind."

  23. DrBobK
    Headmaster

    Human Brain Project

    The 1+ Billion Euro 10-year EU Human Brain Project, which has similar aims and much, much more academic input, has not, by most accounts, been a glittering success. As lots of other people have said, we really know very little in detail about how interactions between neurons in the brain give rise to useful information processing. We don't know which properties need to be simulated and which are irrelevant. There is certainly a great deal of analog computation involved. Some of this is likely intra-cellular. There are lots of cell types which are not neurons but which may play some role. I am not optimistic. (Professor of Cognitive Neuroscience by trade)

  24. John Sturdy
    FAIL

    Ornithopters!

    This sounds like trying to design an aeroplane by copying bird anatomy.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Biting the hand that feeds IT © 1998–2021