Page 1 of 3 123 LastLast
Results 1 to 10 of 28

Thread: A brain in a jar

  1. Top | #1
    Contributor
    Join Date
    Apr 2002
    Location
    Finland
    Posts
    5,228
    Archived
    6,261
    Total Posts
    11,489
    Rep Power
    78

    A brain in a jar

    Fast forward 50-100 years. You are dead. But science and technology has advanced to a point where it's possible to take a human brain, slice it up, and scan every single brain cell and synapse with enough accuracy that the whole thing can be simulated in a computer. So you wake up inside a simulation as a disembodied being; a brain in a jar.

    The first issue is of course communication with the team of scientists who revived you. Otherwise, how would anyone know the simulation is working? The scientists can of course observe your brains directly, it being digital, and interpret the firings of neurons and brain regions. The simplest way is to train with an EEG measurements prior to your death and agree upon some way to pass messages once you are in the simulation. One very important code to agree before the brain scan is a "safe word", that you can use to tell the scientists that something went terribly wrong and they should shut down the simulation. But you also need some way to receive messages from the outside. This could be achieved by the outsiders simulating electrical charges in certain neurons or nerve endings. You could likewise practice this in advance by having electrodes stuck into your brain or outside your skull. Now you have two-way communication, albeit a bit clumsily.

    Having to talk to outside world via Morse-code like system of brain patterns is good for checking that everything is alright, but it doesn't really sound appealing in the long run. It would be much better to be able to get your old sensory inputs back. And there's no reason you couldn't: the nerve endings are still there, all you need to do is to figure out how to connect them. Let's start with vision. The technology to scan individual cells exists, so the science team might as well go the extra mile to scan your optic nerve right up to the cone and rod cells in your eyeballs. Now it's just a matter of simulating the optics of the eye to project any image onto your simulated retina, therefore making you able to see. At first, it'd be odd because you couldn't move your virtual eyeballs nor focus, but it's better than nothing (and it's not like the image would be out of focus; more likely, your virtual vision wouldn't need adjusting because it would always be in perfect focus). Similar thing could be done with the nerves that go to your ears to give your hearing back.

    Getting the rest of your body back would be possible to do the same way, by scanning the entire nervous system. But it probably wouldn't be done that way. First, it's the cost. If you're one of the first people to be scanned, it's easier to start with just the brain. Scanning rest of the body is just an additional hassle. Second, it's not so difficult to learn how to use a new body. Just think about lifting your arm; then connect the nerves that were activated to a virtual arm that you can see. If the arm movement doesn't feel natural, adjust the parameters until it does. Do the same thing with all the other muscles. And same with inputs: activate certain neurons and observe which part of your body they "feel" like they are coming from. After enough practice you'll have a virtual body that can move and feel just like your old fleshy one; learning to use it is a bit like physical therapy and might even be fun. After all, in the virtual world you won't get hurt if you fall down and you don't have to have full control of your body to be able to move around the virtual space. It'll be at first more like controlling a computer game character.

    Your own personal virtual world would be where you spend most of your time anyway. You could of course set up a robot body to interact with the real world, and for some purposes like talking to flesh-beings that's something you want to do, but mostly sticking with the virtual environment is much more comfortable than lumbering around in the real world with all its restrictions. Virtual space is cheap, and there's no reason you couldn't have an entire mansion or a city all to yourself; all you need is the 3D models of the place. Another reason why you would prefer that over the real world is speed. There is no reason why your brain would have to be simulated in real time: the speed of electro-chemical signals in the brain is fairly low, just about 120 m/s. Compared to speed of light that's 300 million m/s that's peanuts. If we were to build an artificial brain at 1:1 proportions, but with copper wires, it would run about million times faster than a biological brain. With miniaturization and clever optimizations you might be able to run the brain a few more orders of magnitude faster even. There is plenty of room to improve on puny brain tissue; even with a simulation running on generalized hardware, it's not unreasonable to assume a speed-up of several orders of magnitude. Exactly how much depends on how fast (and cheap) the computers can get, how detailed a simulation needs to be, among other things, but let's say that your simulated mind can run on average a hundred times faster than its former biological counterpart. It would be incredibly boring to spend that time in the physical world.

    So what do you do with the time? You could decorate and furnish your virtual surroundings. Play games. Read books and watch movies. Talk with your other simulated friends. Study and learn; a degree in neuroscience might be very useful so you don't have to depend on the scientists on the outside in order to tinker with your simulation parameters. And of course, work. You may be a transcended post-human, but you still have to pay rent. The computing time you use is not too expensive, but the main costs come from the fact that you are dead and legally have no rights whatsoever. You need to have someone on the outside that you trust, preferably family members, to handle your affairs and funds. They need to hire lawyers that know IPR and human rights laws to defend you against possible threats. You need to have backups in multiple countries, and extraordinary security measures to avoid your brain simulation from ending up in wrong hands. Imagine the horror of someone uploading your brain patterns to the dark web, where any psychopath could take it and run it in his personal torture dungeon, for example. Eventually the society will come to grant some rights to the simulated persons, but at first it would prudent not to trust anyone.

    I mentioned friends before. Surely you wouldn't be the only one who is scanned, especially once the technology has been proven to work. Given the costs of the brain scan and the aforementioned running expenses, it'd mostly be wealthy people with interest in transhumanism. The fact that one's brains need to be put through a meat grinder for the transition to happen, there are not that many people willing to do that voluntarily: those who can afford it probably have pretty comfortable lives anyway and are in no hurry to commit suicide. By this time, most forms of cancers and dementia can probably be cured, and human life spans extended, so the most likely candidates are those who die in convenient accidents (no brain damage), those dying of old age or with incurable diseases, and those who are really committed. Slowly but surely the number of simulated persons will grow, and because it's so much faster for them to talk to each other than biological humans, they will form their own communities where they exchange their experiences of the transformation, tips how to configure their virtual bodies and homes, and in general just hang around and play games. Some might have a knack for creating virtual furniture and other content for their fellow sims to enjoy; there might be economic activity that is not tied to the real world anymore. At some point some of the folks will be neuroscientists and other experts who can actually start to improve the simulation and experiment with improving their own brains, but that will take a while.

    As for economy in general, as a simulated being you have certain limitations as to what you can do. AI's had already taken over most routine jobs when you were alive, and just being faster doesn't make you any better than regular computers at driving cars or flying drones. You obviously wouldn't be doing as well as a human in anything that requires fine motor control like acting, singing or prostitution. What that leaves is creative pursuits. You could write articles and books, compose music, and in general create various kinds of informative and entertaining content for the humans. What you might lack in creativity and originality, you make up in productivity: you can work 100 times faster than a biological brain. If you were an author, for example, you could write books just as fast as your fans can read them. And you could still do some jobs that require human empathy, such as psychiatry, albeit remotely, with the distinct advantage of being able to formulate your responses with much more deliberation than a human would. Even in fields like science and technology being able to think fast might help optimize resource usage, even though running experiments and building stuff takes just as long as before. All in all, there are going to be plenty of things for the simulated community to do to make itself useful for the general population, even though in numbers they can't compete yet.

    While each year more and more people would join that community, growth would be excruciatingly slow. After the first years, you would mostly be running at full speed (to not do so would be wasteful), so it takes a long time in subjective time between new arrivals. And anything else. New iPhone that comes out every year, now comes out once a century. You might switch to a faster hardware every month or so, but subjectively that's five to six years. And paradoxically, upgrading your hardware isn't helping: it makes you faster, but everything else slows down. This simulated world is more static than the real world, from subjective experience. Computers are slower than they were when you were alive (depending on when Moore's Law reached its peak, it could take you back several decades), and internet is even more so. Instead of fractions of a second for websites to load, you need to wait minutes. And the virtual world you live in isn't quite as detailed as the best computer games. Even if it is possible to have a pretty convincing, real-time VR for humans, now you have to run that 100 times faster, and the requirements for immersive physics model are higher because you need to stimulate not just visuals, but also the sense of touch for the entire body. In practice, you need to make some compromises in terms of how immersive the virtual environment is, if you want to run your brain at full speed. On the other hand, you can always have higher fidelity experiences by slowing yourself down, but then you are making a trade-off between immersion and time.

    Locality becomes crucial: even when talking to fellow simulations, you are better off being co-located in the same server room than being in different cities. And talking to humans is a chore onto itself, because getting any response will take hours instead of minutes. Roughly speaking, a ten-minute chat with a human takes a full day now: 16.5 hours of subjective time and 7.5 hours of sleep (even a simulated brain needs sleep). The upside is that you can do other things while waiting for replies. If you are impatient, you can always slow down, but it's probably better to learn patience than waste time. What you could do is make a copy of yourself to slow down, and later catch up with. In fact, you could do this with other boring jobs. If you need to write a book, just split yourself so that one copy writes a chapter, and you can finish in a fraction of the time. But what do you do with the copies once they are done? If you keep them alive, you would have to split your funds and other possessions with them. Deleting them is one option, but then you can't learn from their experiences. Ideally you would want to be able to merge the copies back into your own brain, not unlike merging files in a version control system, but the algorithms to do that need to be developed first and that takes both time and brave volunteers who are willing to risk messing their brains. So deletion is probably the way to go. It's not as radical as one might think: you already died once, so doing it again is not such a big deal. You lose all your experiences from the time you copied yourself, but will be confident that there is another you that benefited from your work and will continue to live on.

    What about the future of these simulated beings? The community grows slowly, and it will take a long time to overtake humans. There might be a surge if countries of the world decided that simulated brains are the future and they need to upload as many people as they can in order to maintain cultural presence in the future. Some will make copies of themselves and eventually most simulated beings might actually be copies rather than originals: even if most wouldn't do it, there are going to be some who have no qualms about making thousands of copies of themselves for whatever purpose. And I'm sure that at some point the sims will start thinking about how to have children, as that is a pretty fundamental human desire. Killing and scanning a baby is abhorrent so that is not going to happen. Some might experiment with creating blank brains, that could be raised entirely in the virtual reality. But that, like merging copies, takes a long time to develop. Same goes for extending the human brain in ways that biology would restrict, such as increasing the number of neurons or adding new senses. There are some smaller optimizations that could happen easily: for example, you might not need to simulate every ion passing through the neuron cell walls, a statistical estimate could be fine. And you might not need to simulate synapse growth for every cycle. Those optimizations that just increase the efficiency of the simulation without losing fidelity will be taken into use by almost everyone, but some changes that actually alter how the brains work are probably going to be used just by some who want to experiment. But anyway there is some constant, albeit slow progress that takes the simulated community farther away from baseline humanity.

    Consider this: after a few years from the first uploads, there will be people who have lived longer in the simulation than they did in the real world. In a decade, there are people who have lived longer in simulation than any biological human being has ever lived. In less than a century, the virtual community will have subjectively existed longer than the entire recorded human history. Of course, you and the rest of them will form their own culture, and will feel increasingly disjoint from their origins. A question may be posed, do we really need billions of slow, organic humans wasting space and resources on Earth when you could run them in computers with fraction of the cost?

  2. Top | #2
    Contributor
    Join Date
    Apr 2002
    Location
    Finland
    Posts
    5,228
    Archived
    6,261
    Total Posts
    11,489
    Rep Power
    78
    What other fun and interesting details does living your life as a simulated brain entail?

  3. Top | #3
    Contributor
    Join Date
    Sep 2004
    Location
    California
    Posts
    5,240
    Archived
    4,797
    Total Posts
    10,037
    Rep Power
    69
    This is pretty much the premise of the Amazon Prime TV series Upload.

  4. Top | #4
    Contributor
    Join Date
    Sep 2004
    Location
    California
    Posts
    5,240
    Archived
    4,797
    Total Posts
    10,037
    Rep Power
    69
    Quote Originally Posted by Jayjay View Post
    What other fun and interesting details does living your life as a simulated brain entail?
    You wouldn't have to be in top physical condition to become an astronaut. A simulated mind would be the ideal space explorer, since you could spend the boring years between planets with the simulation suspended, and you could beam backup files of yourself to Earth for storage on a regular schedule, so when you get yourself killed doing something dangerous, you're fine.

    I think your scenario of living a hundred times faster than normal people is unrealistic, except maybe for lab experiments by the technology developers. It only works that fast if you run the simulation on billions of processors, and why would you? It would be a lot cheaper to time-share the neurons on only millions of processors so the simulated minds can interact with bio-humans at normal speed; what would justify the extra expense of more massive parallelism for the average uploaded dead guy?

  5. Top | #5
    Veteran Member Lion IRC's Avatar
    Join Date
    Feb 2016
    Posts
    4,638
    Rep Power
    25
    Quote Originally Posted by Bomb#20 View Post
    This is pretty much the premise of the Amazon Prime TV series Upload.
    Yep.
    Also similar theme;
    Altered Carbon
    https://g.co/kgs/D2ENnH

    ALTERED CARBON is set in a future where consciousness is digitized and stored in cortical stacks implanted in the spine, allowing humans to survive physical death by having their memories and consciousness...

  6. Top | #6
    Veteran Member Lion IRC's Avatar
    Join Date
    Feb 2016
    Posts
    4,638
    Rep Power
    25
    Quote Originally Posted by Jayjay View Post
    What other fun and interesting details does living your life as a simulated brain entail?
    The conscious entity could or could not freely leave the jar???

  7. Top | #7
    Super Moderator
    Join Date
    Sep 2000
    Location
    Nevada
    Posts
    32,983
    Archived
    96,752
    Total Posts
    129,735
    Rep Power
    100
    Upload is a common Sci-Fi idea.

    Note that in addition to the creative pursuits there's another job, albeit in limited numbers: Operator of distant robotic equipment. Consider how little the Mars rovers actually drive due to the need to check everything carefully and the lightspeed lag in doing so. How much better would they be if there was actually a mind on board. At the end of the mission you just beam the mind home, an awful lot cheaper than bringing the crew home.

    Note that you can always slow down the speed of the simulation if need be.

  8. Top | #8
    Contributor
    Join Date
    Apr 2002
    Location
    Finland
    Posts
    5,228
    Archived
    6,261
    Total Posts
    11,489
    Rep Power
    78
    Quote Originally Posted by Bomb#20 View Post
    I think your scenario of living a hundred times faster than normal people is unrealistic, except maybe for lab experiments by the technology developers. It only works that fast if you run the simulation on billions of processors, and why would you? It would be a lot cheaper to time-share the neurons on only millions of processors so the simulated minds can interact with bio-humans at normal speed; what would justify the extra expense of more massive parallelism for the average uploaded dead guy?
    There are a lot of unknowns how fast the computers can get. I think the biggest hurdle is scanning technology, not computers. Moore's law may be slowing down, but there are a lot of things we can still do: new materials, nanosheet transistors, stacking transistors in 3D, and massive parallelization. Not sure how far we can get. But whatever the endpoint, I think that it's unlikely that the economically feasible computing capacity will be in the same order of magnitude as is required for a brain simulation. But also, I do think the first volunteers will be millionaires and not regular joes precisely because of the cost. After it becomes more commonplace, there are going to be specialized hardware that is optimized for the specific task of brain simulation. The same thing that happened with mining bitcoin: it started with general purpose CPUs, then moved to general purpose but parallel GPUs, and finally to custom ASICs.

    There is also room for software optimization after you have an actual person in the simulation. A dog can't tell you if its brain is working correctly, but a person can. You can use that introspection to figure out which parts of the neuron need to be simulated more accurately, and which can be just approximated.

    I suppose the answer could also be a matter of economics. If you earn $100 an hour, adding 10% more computing power earns you extra $10/h. If the cost of 10% more cores is more than that, then it's not worth it. Roughly speaking; running faster also gives you more spare time which you may value differently from your job productivity.

  9. Top | #9
    Contributor
    Join Date
    Apr 2002
    Location
    Finland
    Posts
    5,228
    Archived
    6,261
    Total Posts
    11,489
    Rep Power
    78
    Quote Originally Posted by Loren Pechtel View Post
    Upload is a common Sci-Fi idea.

    Note that in addition to the creative pursuits there's another job, albeit in limited numbers: Operator of distant robotic equipment. Consider how little the Mars rovers actually drive due to the need to check everything carefully and the lightspeed lag in doing so. How much better would they be if there was actually a mind on board. At the end of the mission you just beam the mind home, an awful lot cheaper than bringing the crew home.

    Note that you can always slow down the speed of the simulation if need be.
    There are a bunch of things that you can also do with "dumb" AI. Driving a Mars rover sounds like a job for specialized AI system, rather than a human. Of course, there are other jobs in Mars and elsewhere in solar system that may require more human ingenuity, and maybe simply visiting Mars is a valuable experience.

    In general I totally agree with you... space is a harsh place for humans. We'll go there for sure but in the long run it's too inhospitable for anyone except the adventurers. Uploaded humans though wouldn't care as long as they have a power source and reliable hardware to run on.

  10. Top | #10
    Super Moderator
    Join Date
    Sep 2000
    Location
    Nevada
    Posts
    32,983
    Archived
    96,752
    Total Posts
    129,735
    Rep Power
    100
    Quote Originally Posted by Jayjay View Post
    Quote Originally Posted by Bomb#20 View Post
    I think your scenario of living a hundred times faster than normal people is unrealistic, except maybe for lab experiments by the technology developers. It only works that fast if you run the simulation on billions of processors, and why would you? It would be a lot cheaper to time-share the neurons on only millions of processors so the simulated minds can interact with bio-humans at normal speed; what would justify the extra expense of more massive parallelism for the average uploaded dead guy?
    There are a lot of unknowns how fast the computers can get. I think the biggest hurdle is scanning technology, not computers. Moore's law may be slowing down, but there are a lot of things we can still do: new materials, nanosheet transistors, stacking transistors in 3D, and massive parallelization. Not sure how far we can get. But whatever the endpoint, I think that it's unlikely that the economically feasible computing capacity will be in the same order of magnitude as is required for a brain simulation. But also, I do think the first volunteers will be millionaires and not regular joes precisely because of the cost. After it becomes more commonplace, there are going to be specialized hardware that is optimized for the specific task of brain simulation. The same thing that happened with mining bitcoin: it started with general purpose CPUs, then moved to general purpose but parallel GPUs, and finally to custom ASICs.
    CPU speed doesn't really matter--the human brain has a truly tiny processing speed. Human brains are massively parallel and thus a simulation of a human brain can also be massively parallel. You can throw as many CPUs at it as needed, there's no doubt we could construct a computer with human-level processing power now. (It would be mighty expensive, but not enough to deter those near the top of the Forbes list.) The sticking point is getting the information out of the brain and into the computer.

    Note, furthermore, that you don't really need full human processing power for the first uploads. Yes, the uploaded person would think very slowly--but that's better than death and in time faster machines would become available.

    There is also room for software optimization after you have an actual person in the simulation. A dog can't tell you if its brain is working correctly, but a person can. You can use that introspection to figure out which parts of the neuron need to be simulated more accurately, and which can be just approximated.

    I suppose the answer could also be a matter of economics. If you earn $100 an hour, adding 10% more computing power earns you extra $10/h. If the cost of 10% more cores is more than that, then it's not worth it. Roughly speaking; running faster also gives you more spare time which you may value differently from your job productivity.
    I'm not sure the economics quite work that way--there's little in the way of an external clock setting your work/leisure balance once you're uploaded. For most people most of their money goes to upkeep and for most people they need to work when everyone else is working. In a computer your upkeep costs are lower and in the creative fields working when others are working is much less of an issue--especially if you're living to different clocks and engage in little realtime communication with your clients and possibly coworkers.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •