Can people upload their consciousnesses to computers?

Question

The idea of uploading one's personality and consciousness to a computer is a semi-common science fiction trope. It has been suggested as a way to essential immortality.

In a story I'm working on, an entire society has transferred itself from brains to computers. They're on a spaceship, heading to a new world, fleeing from an apocalypse on their home planet. Putting themselves on computers is much more efficient than the alternative.

Can they upload their consciousness to computers, keep the whole thing running smoothly, and hold them in a state such that they might be transferable to organic bodies at a later state (e.g. upon finding a habitable planet)?

Some specifics:

• The society is composed of about 3 million people
• The spaceship is roughly the size of a Super Star Destroyer (from Star Wars)
• The journey will take many, many years - quite possibly several centuries
• The technology level is roughly within 50 or so years of today's
• All answers must be based in hard-science

Answer 1

It depends what you mean by 'conciousness.'

Unfortunately, we aren't really sure what exactly consciousness is, or how it can be defined.

One one hand, with sufficient advances in technology, it may be possible to create a computer that can act exactly like a human, and to map a human's brain. If we can do this, we could take a crew of humans and map their brains to a computer. If consciousness is an emergent phenomenon based on the structure and activity of though, then this is uploading consciousness. Assuming that we get the process right, the computer would, at the very least, view itself as being conscious, and being an uploaded version of the consciousness of the uploadee.

Assuming that we can 3D-print a brain, which seems to be the logical endpoint of current research in printing tissues and organs, we could similarly print a stored copy of the uploadee into a new brain at the destination.

Of course, a philosopher may well claim that we've merely created a copy of the consciousness or mind in question. A priest of many religions would certainly claim as such. If we can scan a brain in a non-terminal manner, it may well be that the person being uploaded will still exist in a fleshy physical body, which can point to the computer and say, "That's not my counsiousness." And they may be right, since we don't know what consciousness is.

For the purposes of leaving a planet and re-establishing the race on a different planet on the other side of a multi-century journey, a more pertinent question may be "does it matter"? Regardless of how we go about uploading a consciousness, and regardless of whether or not it truly represents the copied being, the minds we are storing in the space ship are functionally the same. They have the same memories and can do the same things as the original, can be printed back into a physical body at the end of the journey.

Technologically, of course, being able to scan and copy a brain, as well as being able to create a computer than can function in the same manner as a brain is a stretch for '50 years in the future'.

Then again, so is creating a several kilometer long spaceship capable of interstellar flight. If your heroes have accomplished that, I suspect they have the resources and capabilities necessary for copying a brain.

Answer 2

I've posted a detailed sumary of brain/computer correspondence with timeline and my projections on another Answer before.

Look at the graph. 50 years is a no-brainer (pun intended) for being able to simulate a brain with a far greater level of detail then needed given the understanding we already have.

The study of the brain is funded to a pittiful level compared to how much society spends on everything in total. If it became a priority and everybody put in the price of a movie ticket, it would be done in a few (single digit) years, mainly due to time needed to fabricate new designs of hardware.

As for "duplicating", the only way of scanning known now is destructive. We don't know if it's physically possible to scan the necessary level of detail using non-invasive techniques. Nanotechnology might be applied to crawl around inside and make maps, or take the tissue apart and put it back together again.

That could be an interesting part of the story: destructive scanning was used to depart, and they will work on the problem of building a replacement organic brain while en route. They have time and nothing better to do, but few physical resources and rationed processing power. Should poets be left on the shelf to devote more run-time to the biologists?

Answer 3

It depends greatly on philosophy.

I will consider only the viewpoint of structural physicalism, not any dualistic or idealistic answers. The rationale for this is simple: if there is a metaphysical mind, the phrasing of your question makes no sense. If there is a metaphysical mind associated with consciousness, we wouldn't be using the hard-science tag now, would we!

The fundamental challenge for such a transfer is that the brain is a living organ. As such, it doesn't stop moving nor growing, mostly through the creation and destruction of synapses, and the electrical activity within the brain. The brain is not "quiescent," which is the technical term in simulated automata for a structure which holds still. Thus, while you are mapping the brain, it is changing. You would likely end up with tearing like effects like those seen in automated panorama stitching. Sometimes the stitching works acceptably. Other times the results are, well, disconcerting. This issue would be a major challenge for 3d printing a brain. While other organs are more forgiving with their internal structure, we are rather dependent on the structure of the brain to be "right."

There may be solutions involving extremely low temperatures. If you can induce brain death (no electronic activity), chill the brain down to very low temperatures (to restrict synaptic growth), and sample it, you could reverse the process to print the brain. However, the psychological effects of being rendered brain dead are hard to grapple with.

I would consider a different option, which I would call "budding." If you believe a computer is capable of holding onto a human consciousness, then you believe it has the raw building blocks needed to form something human-like in hardware/software. Consider a case where you create a region of a computer for "use" by a human. Within this region is the building blocks for whatever soul is believed in. The human is then encouraged to build this empty soul into a helpful conscious entity.

This initial phase gives an opportunity to make an entity which is reasonably relatable to by the individual. Eventually it should act like an AI. Now, one opens up the AI to directly influence the world (such as giving it access to the Internet). Very soon, you end up with a relationship that should look similar to a working dog, except with a human-grade intelligence within the computer rather than a dog-like one.

With enough training, a human could be trained to treat the human body and the AI mind as part of one entity, just as those with prosthetics learn to treat the prosthetic as part of their own self. Now half of the combined mind can be put in a computer, simply by breaking the link. However, we want more than that, so lets train the computer to become more and more like the human, until the human can no longer distinguish the difference between the human mind and the AI mind.

At this point, we call for "divorce." This human/computer entity agrees that it will undergo a divorce process where the link will be servered, and whatever stays on the human side will be human. Whatever stays on the computer side will be computer.

You now have a cloning problem. You have two individuals that lay claim to the "consciousness" that was the human/computer. Hopefully there was an agreement ahead of time... maybe Bob becomes BobH and BobC, just to avoid ambiguity. Get all of the legalese in place.

Now you have two entities which both feel they are the natural consciousness of Bob, one of them in a computer that you can send far away.

At the destination, the process can be reversed, generating an organic body which is the extension of BobC. At some point this computer/human body also goes through a divorce, creating BobCH and BobCC. BobCH is in an organic body, and believes it is the naturally following conciousness of BobC, which believes it is the naturally following conciousness of Bob.

Now Bob and BobCH are probably different in many ways. Anyone who one would probably instantly realize that the other is not the same. However, given your story has Bob dying in an apocalypse, there is nothing to compare BobCH against. BobCH claims to be the conciousness of Bob, and there is no evidence to challenge this.

In fact, if BobCC is willing to drop the claim to the conciousness of Bob (either as an understanding of the purpose of computer encoding a human in the first place, or through forcefully erasing BobCC), BobCH can even elect to drop the clarifiers, and assume the name Bob. Nobody would be the wiser.

Answer 4

I think it may be possible, so long as the inputs, outputs, and internal variables of certain neurons or lobes of the brain can be narrowed down and replaced with technology that can perfectly accommodate the replaced pieces. Most likely through Nanotechnology.

Some preliminary numbers on all of this; The most convenient method of deploying nanobots is to swallow a pill full of them, but then you still need to cross the blood-brain barrier. The pore-size is typically <1nm, but with a bit of cellular engineering (carrier-mediated transport), we can increase that to a diameter of several nm. [Modern SSDs put 10 terabytes on a 63,500,000nm drive, which is 173kb per nm... And that's enough for a simple OS. And in 50 years time, who knows how much that will improve.

The current bottleneck for this is computer processors. The transistor we assume to be the smallest possible is still a massive 5nm, and processors are a network of transistors; even the simple Intel 4004 had 2,300 (which would be 11,500nm if implemented with 5nm transistors).

Though this boundary could perhaps be surpassed if some of the logic gates for the ALUs were made out of diodes instead of transistors, it is more likely if the Nanobots operate as an emergent distributed processing network; each one communicating with others nearby in order compute, sort of like the human brain itself.

For purposes of scale, Neurons themselves are between 3000nm and 18000nm, so plenty of these <10nm nanobots may have to be deployed in order to assure a reliable network and a strong signal. A functional CPU would also need a register bank to keep track of control flow, and an ISA to allow the CPU to 'run software', but that's a feat for engineering to surpass.

Answer 5

I'm going to start with I think that is actually a great way to fly across the cosmos.

However, I don't think uploading 'consciousness' is something we can do with current understanding of the brain. Our first problem to solve would be just to be able to record memories to digital format in such a way that we can 'recall' it. I suspect that will be our first step. More and more people are using things like Google-glass to record their lives, the next step would be to record more. Using external/artificial devices as backup memory. Once we can record and play back memories and experiences we will be on the road to uploading conscienceless. Memories are a rather simple step compared to the next.

It isn't just that you need to have enough processing power, or harddisc/memory to hold the whole personality, memories and experiences, you also need to be able to simulate the interconnectedness of human thought. While we continue to learn more and more about how our brains work, we are still a long way from that.

So the next step would be if we can up load an animal, such as a well trained dog, can we take a cloned copy of the dog 'write' back to it? Will the dog remember it's tricks? Will it act the same?

Writing memories to hardware storing them and then writing them back to wetware is still not allowing the person to 'live' while in the machine, it's just making a 'backup' that hopefully can be stamped onto a new copy, (or even the old body out of cold storage) to let the person continue where they left off.

The last and be far most difficult step is to allow people to actually be conscience while in the machine. I would certainly not volunteer to be the first test subject unless I'm dying from something. The first problem here is we don't really know what consciousness really is. So how can you model something if you don't know what it is?

I suspect that after we start having hardware backups connected to our bodies and networking becomes more ubiquitous experiments will start with 'traveling' through the networks to 'be' somewhere else, a VR type existence. This would help prepare people for actually being separated from their bodies. Most of us today would likely go insane in fairly short order by removing us from our bodies, none of our expected inputs available, our body not telling us our but is sore from sitting, our bladder is full, we're hungry, not to mention sight sound and smell etc. We would somehow need to be able to use external sensors to get input.

So in 50 years I think it is possible we will have the ability to record and play back memories and experiences. Might even be able to share them with others. Which makes me think a whole new wave of porn would be born. You wouldn't just 'watch' you could 'experience' things. Many other interesting things could be done, both good and bad. But I think we'll be short of a full upload download yet. Not for lack of equipment capabilities, but lack of understanding how everything interacts and the ability to program for such a monumental task. Though I expect we'll have Conscious AI's which would dramatically help us understand what we need to do.

Answer 6

Absolutely. The human brain is nothing but a chemical computer, after all, and it's this chemical computer that gives rise to the mind. For me, the problem of causality is a watertight argument against any notions of duality. Once one embraces monism, I feel that the possibility of transferring one's consciousness to other substrates becomes obvious.

Since dead organic matter as we know it now tends to spoil over the course of centuries, I think an intermediate step may be necessary if we're going to realistically make this happen within the next 50 years. I think mechanical bodies would be a better bet in terms of durability and longevity in storage. And if things end up taking longer to develop than planned (they almost always do, it seems), the mechanical bodies wouldn't necessarily have to house the consciousnesses controlling them. They could be "thin clients" (or "thin avatars" in this case, I guess) for the consciousnesses still in the megalithic computer on the "lifeboat." That would allow the survivors to then build organic bodies using what's available on whatever planet they end up on.

Of course, cryogenics is another possibility. But that would be very energy-intensive compared to just making hyper-durable robot bodies that would remain switched off for most of the trip. My money would be on the robot bodies. Totally a possibility, because space is cold and most of the trip would be done with solar sails or something else of similar efficiency, as @bowlturner pointed out in the comments below.

Note that you wouldn't need 3 million robot bodies, of course. You'd just need enough to allow everyone who needs to interact with the physical world to be able to do so. More is better to a point, but 3 million seems like it would be excessive.

For technological paths that would likely lead to the vision you have here, look at neuromorphic computing and biocomputing. Again, due to the fragility of organic matter, I suspect biocomputing may be a dead end for transporting consciousness over the span of centuries. However, I also suspect it would prove to be very useful when it comes time to start creating organic bodies on the new home planet.

Not sure if this has enough hard science for you, but hopefully it's helpful all the same!

Answer 7

I can't do the hard-science for your other two questions, so I'll try to just focus on this one.

keep the whole thing running smoothly

Note: I've made a ton of estimates and assumptions in this answer. There are obviously many different ways to tackle this problem, and this is only one. There are also many, many other considerations to make in regards to the scenario, but I have indiscriminately hand-waved them away for now.

First of all, let's try to figure out how much storage space you'll need to hold your civilization. Estimates for how much a human brain can hold vary from 1 terabyte to 2.5 petabytes. Let's just take the high end of that estimate and work with that.

You're going to want some kind of redundancy as well. There are different strategies for redundancy, especially considering these are people we're talking about, but for simplicity, let's go with a RAID 10 approach.

$$ 2.5\ petabytes * 4\ drives * 3\ 000\ 000\ brains = 30\ zettabytes $$

Now let's consider what medium of cold storage to use. Unless you specify otherwise, there should be no reason why these brains need to be conscious during the trip. From the passenger's perspective, the passage of time should be imperceptible from the time their consciousness is loaded into storage, to the time their new bodies are created. So let's use magnetic tape for our long-term storage. In my opinion, there really is no better way to keep things running smoothly for your scenario.

The latest generation of LTO Ultrium (the standard form factor of Linear Tape-Open magnetic tape drives), LTO-6, has a raw data capacity of $2.5\ terabytes$. Currently, future generations of LTO are already planned, all the way up to LTO-10. So, let's go with LTO-10 as our storage medium.

LTO-10 has a planned raw capacity of $48\ TB$. At a compression ratio of $2.5:1$, we should be able to store $120\ TB$ on one tape.

$$ \frac{30\ ZB\ of\ storage\ needed}{120\ TB\ per\ tape} = 250\ 000\ 000\ tapes $$

Using this product's specifications as the basis for what a future-gen's specs might be, the mass of all those tapes would be $250\ 000\ 000 * 0.27\ kg = 67\ 500\ 000\ kg$, or about times the mass of the Titanic. The volume for all those tapes would be $11.3\ cm * 2.79\ cm * 11.1\ cm * 250\ 000\ 000 = 87487\ m^3$, or roughly 0.4 times the volume of the Hindenburg. As of writing this, you can buy this LTO-6 tape at 10 for 320.00 USD, or $\$32.00 * 250\ 000\ 000 = \$8\ billion$, or the value of Spotify. According to these numbers, I think your Super Star Destroyer-size ship is probably a little excessive.

Of course, your ship and its payload will have to be maintained and piloted somehow. I don't know if you want to allow robots to perform the routine maintenance and piloting, or have some small group of humans with bodies awake throughout the journey...that would add a layer of complexity to your ship design, for sure. Or you could have some ungodly combination, where you have a few human consciousness' plugged directly into the shipboard controls and sensors, able to maintain the ship. That actually sounds like a pretty awesome plot setup.

The good thing about these tapes is that they have very easy requirements. It looks like the ship library they are to be stored in should be temperature- and humidity-controlled at ranges between 5° to 23° C, and 20 to 60%, respectively. And that's about it. The rest of your ship and its energy is dedicated to just being a ship.

Let's say we use these bad boys for our data library. Of course, we'd probably have a custom build in our ship, but I'm just using these as a basis for comparison of specs. Also, I'm assuming data transfer rates are static over the next 50 years. We'll need $2\ 500\ 000 \div 12\ 006 = 20\ 823$ of them to hold all our brain tapes. The bad thing: these are probably crazy expensive. I couldn't find a number, but I can only imagine. The good thing is the R/W speed. If we were to use a single drive to read out all of the tapes it would take:

$$ \frac{30\ ZB}{400\ MBps} = 2\ 377\ 000\ years $$

Since we were RAID 10 redundant, we can go ahead and cut this number in half on the read side, when we get to our new home planet. In order to cut the total read/write times to within 1 to 2 years, we'd need about 1.5 million tape decks. At about \$5,500 per drive, they would cost us about another \$8 billion.

But with our fancy tape libraries and reading decks, we would have all the reading done in

$$ \frac{30\ ZB}{138\ TB/hr * 20\ 823\ libraries} = 435\ days $$

Not too bad, considering.

Answer 8

Upload, yes. The data storage requirements are extreme but not beyond what T+50y tech should be able to provide in a crash-priority program such as this. We don't need to understand the brain, T+50y tech should also provide enough compute power to emulate the brain at realtime speeds. (If Moore's Law holds we should have supercomputers at this power level at T+20y. Since extreme parallelism is acceptable I can't imagine this target not being hit.)

The other half of your question is an unknown, though. We simply don't have enough understanding of the brain to know if we could build a functioning brain with memories or not at T+50y. However, this would not be a showstopper--in a do-or-die situation like this the ship would fly anyway even if they don't have any download system in place. Bring a whole bunch of fertilized eggs in cold storage and bring the computer-encoded DNA of even more people, along with the coding for everything that makes up a human egg.

The thing is the one thing you do have plenty of at that point is time. Live in the computers until such time as it becomes possible to download--not that I'm at all sure that people would choose that route. I strongly suspect you would end up with a society of computer-housed minds and tele-operated robots for when something hands-on needed to be done.

Answer 9

If you can upload a human mind into a computing device, then you've essentially reached the point of technology singularity.

A space going vessel is the logical place to keep the computers in which the minds live, but why seek out planets or recreate organic bodies at the other end? Minds living inside a computer are free to generate any simulated environment they desire, including any simulated body they might like to present as their avatar. We already have primitive examples of that in virtual reality environments such as Second Life. What if our computing power was such that we could host full human minds in realistic but user-determined environments?

You may find that your minds, accustomed to the God-like near-instant fulfillment of their needs and desires at-will, may be unwilling to leave their simulacrum to return to physical reality.

The minds become the spaceship. They don't need a planet to live on. Having left meatspace behind, they live in cyberspace full time. A dynamic, digital reality continuously shaped and shaping the minds within.

The practical acquisition of raw materials for energy and propulsion to keep the mindship going is achievable without having to set organic foot on any planet or asteroid by using remotely operated drones or direct scooping of space-bourne gasses, so why bother with fragile, clumsy organic bodies? Physical space, improvements to the ship, computing power, and the continued exploration science can similarly be explored using simulation and drones.

The definition of a singularity is that you can't see beyond its surface. Likewise, with our technological singularity, we can see the surface, the starting point, but our predictions become increasingly hazy the further past that point we go. However, given the natural human desire to grow, it's likely such minds would desire additional computing resources. More memory. More processing power. Improved mind design. Improved psychology. With ever increased (and faster) intelligence, these trans-human minds would soon become both unrecognisable and unfathomable to their organic ancestors, and a new race, existing simultaneously both in space and in cyberspace, is born.