Pages

Tuesday, March 15, 2011

Transcending the Human Form: Robot Bodies VS Virtual Proxies

My last post about AI was perhaps overly simplistic, especially in regards to the complexity of the brain, but I had other points to make. I hope to see the emergence of strong AI in my lifetime, but it's difficult to guess what that will require, much less its consequences. One thing I am much more certain about is the expansion of humanity's capabilities through computers and cybernetics.

Our brains interface with the world through our bodies. It's an ancient interface that has served us well, but it is increasingly restrictive in our evolving high-tech world. Our bodies confine and even imprison us, especially when something goes wrong with them. By improving the body, we enhance the way we communicate, our ability to access and manipulate information, and our ability to deal with the physical world.

Cybernetics and bio-engineering will augment our natural bodies, but what about adopting new bodies entirely? Currently we have no alternatives, but that may soon change. Placing our brains inside robot bodies is one appealing way of transcending the limitations of our flesh. Who doesn't think robot bodies are cool? They would have better senses, durability, endurance, and a finer ability to manipulate the physical environment than any natural body.

But I believe that the true future of humanity is not in the physical world at all, but in the virtual. The full potential of the human brain can only truly unfold in virtual reality. VR offers greater freedom than the physical world in almost every way. Inhabiting virtual bodies, or "proxies" as I call them in Living Outside, our brains will finally be able to express themselves to their ultimate extent.

I want to compare robots with proxies to demonstrate that simulated living is superior to even the best physical existence. But first I want to clarify a few issues.

The Brain Interface
To inhabit any synthetic bodies with full immersion (all the senses and motor controls of the natural body), will require brain implants. These implants will either be integrated into the brain itself, such as with the somatosensory center and motor cortex, or connected to the brain's input/ouput connections– the 12 pairs of cranial nerves and the spinal cord.

The input/output option will likely be accomplished long before cyberization of the brain, and will be sufficient to allow the brain to inhabit and control a robot or virtual proxy with our full complement of senses. Cyberizing these connections will be much less invasive than fiddling with the brain itself, and significantly more straightforward. Axonal frequencies will be far easier to decode and stimulate than complex cognitive functions. Incidentally, the bandwidth of the brain's connections to the body is somewhere around 1 gigabyte/second, which is very manageable.

However it's done, the signals sent from and to a robot or virtual body will be electronic in nature, translated from the sensors of the robot, or the simulated senses of the proxy, into nerve impulses which the brain can interpret, and vice versa. This means that if your brain is hooked up to a robot, there is no reason why you can't route it to a proxy instead. Conversely, if your brain can be hooked up to a virtual body, you should be able to telepresence into a robot.

I want to note that telepresence into a nearby robot should work about as well as actually having your brain inside it. This is because neural signals move extremely slowly compared to the speed of electronics, so the brain shouldn't notice the distance. If all you want to do is run around in a robot, there should be no good reason to remove your brain from your birth body to do so, and you should be able to effectively "become" a robot by wirelessly redirecting the input/output pathways of your nervous system.

The Brain Vat Option
Brain vats are another option for handling a removed brain. They have negative connotations, due to associations with mad scientists and existential thought experiments. However, I think that brain vats are a more likely scenario than putting brains inside moving robot bodies. Stationary containers designed to protect and sustain cyberized brains are safer, more economical, and less technically challenging than robot bodies. They're also better in case something goes wrong. Imagine how easy brain surgery would be on a brain in a vat.

Mobile robot bodies with onboard brains will require dependable life support systems, and well protected brain compartments. Facilities housing brains would have economy of scale, possibly housing hundreds or thousands of other brains. And you can always telepresence into a robot from a brain vat if you really want to run around in one. You can take bigger risks with such a robot than if it had your brain inside it.

Those things established, I will now compare proxy and robot bodies to show that proxies are superior in most respects:


Cost: Advantage Proxies
For either option, there would be an initial cost of implanting the neural interface. Beyond this, a robot which could physically sustain a brain would obviously be extremely expensive, though its costs would go down with time. Robots also require parts, maintenance, upgrades, repairs, electricity, and storage space. Acquiring additional robot bodies will bring their own costs. I have a hard time imagining robots like this becoming cheap, but who knows?

Proxies, on the other hand, would be cheap, if not free, beyond the initial cost of neural implants. You could potentially acquire thousands of different proxies, and modify them however you want, for little or no additional cost. Virtual proxies and environments will eventually require negligible electricity costs.

Immersive virtual reality requires much less technological advancement than robots, and will occur sooner. Inhabitable virtual proxies are much more likely to be developed, period. If you can hook a brain up to a robot, virtual reality is a certainty. But the advent of virtual reality says much less about the feasibility of inhabitable robots. Past the brain implants for full immersion, proxies require only the development of realistic virtual worlds, which are being intensely pursued in movies and games, and the simulated bodies to explore them. I don't know how likely a brainbot is in my lifetime, but I'm fairly certain I'll see full immersion virtual reality if I live another 60 years.


Convenience: Advantage Proxies
Using proxies, you could teleport instantly to any physical place that is virtualized. That is, any place that has sufficient sensors to allow it to be simulated for the purposes of a virtual proxy. Robots could go to any space they could travel to, including places with no sensors. However, they can't teleport, and so it will take time and energy to move them from one place to another. I can't underscore enough the convenience and power of being able to instantly appear in countless locations all over the world.

Proxies never have to be cleaned, recharged, or fine tuned. If something goes wrong with them, it's easy to restore or replace them.


Flight: Advantage Proxies
Proxies could fly like nothing else! The experience of flying as a proxy will eventually become realistic, and could be enhanced from there. Depending on the rules of their virtual environment, proxies could hover or fly anytime they wanted. They could rocket to 700 mph, feeling the extreme acceleration and air pressure, and stop on a dime. They could zip through the sky like Superman. They could dart around a crowded room in a blur without worrying about hurting anyone.

Flying robots won't be nearly as free. They'll be expensive and require lots of energy. Even with a jetpack, robots will never be able to maneuver and accelerate like a proxy, and so the experience they offer will be less thrilling. The only extra thrill from putting your brain inside a flying robot would be from the danger to your own life. I believe that a perfectly simulated flying proxy experience would hold a similar sense of danger for proxy users. A realistic simulated experience should trick a user's brain, even if they are perfectly aware of being in no danger, even better than roller coasters do.

In any case, there would be real danger in flying in a robot body, as a physical crashlanding would at least give your brain a good jolt, even with good brain suspension. Good luck if your life support system gets damaged and you're stuck hanging off the side of a building. Putting my brain in a flying robot is not my idea of a good time, and telepresencing into a flying robot would be a waste of fuel. The sensations of flight produced by the sensors of a robot could be copied and simulated virtually. It would eventually be an identical experience to the user, so why not just use a proxy? Also, flying robots are dangerous to everyone around and underneath them, and their flight would have to be strictly controlled, especially in populated areas.


Interacting with physical environments: Mixed, Robots Overall
Robot bodies will have some advantages when interacting with a physical environment. Of course, first they have to get to the location, which again, costs energy and time. A user could telepresence into a robot at the desired location, of course, though they will have to either own the robot, or make some arrangement to borrow the robot. I should note that there are already an increasing number of robots that are controlled remotely to accomplish specific tasks, used in labs, construction, for surgery, chatting, and many other activities.

Robots can manipulate a physical environment, something a proxy couldn't do by itself. A robot can pick up a physical object and look underneath it. Proxies could interact with virtualized objects and environments, but while this experience could be as real to the proxy user as if they were really there, their actions would make no physical change without assistance from people or machines present at that location.

A robot's sensors would allow its user greater control in examining an environment. A physical space with sufficient sensors could be simulated to create a nearly identical experience for a proxy user, though there could be blindspots. A robot would be more consistent in the experience it delivered, but it would potentially have more limitations. A well constructed virtualized environment would allow a proxy to explore and manipulate objects much more easily than a robot could. Proxies could check out something on a high shelf, sit on a chandelier and tinkle its crystals, shrink down to hide under a couch, or examine locked rooms.

To mitigate some of their disadvantages, proxy users might remotely control special surveillance bots designed to move around a physical environment to better virtualize it. The "bugs" in Living Outside, Episode 3 are examples of these.

Robots could possibly eat, drink, and smoke if they had synthetic digestive tracts and lungs, seemingly necessary to support a brain in the first place. This possibility is interesting, but I have no idea when such tech might be developed and become common. Again, brain vats seem more likely.

The consumption of physical goodies is something that a proxy user could only do on their own end, where their brain was located. Proxy users could not drink the liquor in their friend's house, for example. They could have a perfect simulated experience of swallowing it, but it wouldn't get them drunk. Drugs similarly would have to be imbibed in the proxy user's physical space, and friends couldn't share them.


Interacting with physical people: Mixed, Robots Overall
Here's another area where robots have advantages, though not as many as you might think at first. People with sufficient brain implants could simulate the experience of a virtual person's presence. They could feel a proxy's hug, have sex with them, and otherwise interact with them as if they were really there. It would, in essence, be the interaction of two proxies. The extent of this interaction would be greater than that possible with a robot, because of the greater flexibility allowed by the virtual.

People with non-implant augmented reality technologies could interact with a proxy to a degree satisfactory for most purposes. They could see proxies with bionic contacts or feel a hug from a proxy through haptic clothing. Those without personal augmented reality would have a harder time perceiving a proxy, and their interaction would be limited. They could see and talk to them through a screen, but they wouldn't be able to touch them. For the unaugmented, interaction with robots would be superior. Though I wonder how many unaugmented people there will be by this time frame.

Intimate physical interaction with an unimplanted person would require robots. Proxies couldn't give a physical massage, or have physical sex. Telepresence into a robot that was relatively nearby would confer these same advantages without the need to remove your brain from your organic body, although lag would become an issue over longer distances. Lag could also be an issue for proxies.


Assorted things: Advantage Proxies
So many things that are completely trivial to implement for a virtual body would be quite difficult, expensive, or prohibitively impractical for a robot body.

Even if a robot body could fly, could it hover safely through a crowd? Could it stand on the ceiling and walk around upside down? Could it grow giant, or shrink to doll size? Change form and appearance at will? Turn completely invisible or incorporeal? Phase through walls? Taste with its knees? Blow up? What about turning into a tentacle monster, or copying someone's appearance? Could it teleport to China on a whim?

For a proxy, upgrades and adding components are a simple matter of a software update or tinkering with settings. Upgrading robots will not be so easy.


Interacting with virtual environments: Complete Advantage Proxies
A robot can't go into virtual environments, except by proxy. And virtual environments are seriously where the future is at. Their scope, quality, and diversity will far exceed anything found in the physical world. They will be more beautiful, more engaging, and far more interactive than anything we could build here, at very little cost. I've been exploring these possibilities while writing Living Outside, and they are pretty exciting.

Many people say, "Where is virtual reality? They've been promising it for decades!" I have two responses to that.

First, we already have VR! Look at cutting edge games like Crisis 2. That's virtual reality. The player just isn't immersed in its environment yet. True, it's not as good as people want. Computer technology isn't yet capable of delivering an immersive VR experience. That will require much more intensive processing power to simulate a realistic environment for all the bodily senses. We are only now approaching photorealistic graphics and physics engines of the required sophistication. We've also started working on the tactile elements of virtual objects, such as their texture and weight. We are quickly approaching a time when the average person's computer will be able to simulate realistic, fully immersive worlds, but the hardware isn't there yet.

Second, it will take some effort to create the interface that will completely immerse users in the amazing simulated realities we are creating. Decent audiovisual VR gear is around the corner. Soon, VR capable glasses that also provide untethered augmented reality will have enough utility for people to start paying attention to virtual reality again. Eventually, they will be small, lightweight and stylish. But full immersion VR, featuring senses like touch, acceleration, taste, and smell will require brain implants to be fully convincing. That's much further away, but it's being worked on. The applications of full immersion simulated reality are just too incredible for the necessary tech not to be developed, if it is plausible at all. Just think of the immense market forces that will drive the development of realistic virtual sex, or convincing telepresence.


Conclusion: We All Win (hopefully)!
As I said at the beginning, robots are awesome. I like to imagine that people will get a robot body to encase and protect their brains while they frolic in the liberated expanse of virtual reality. They would be useful for moving your brain in case of emergencies, if nothing else. That sounds like an ideal setup to me. But once people get used to proxies, they're not going to be very interested in robot bodies, except as infrastructure. They might take over one remotely every now and then for the novelty, but the experience will be inferior overall, and I don't think people are going to want to risk their brains by moving them around when they don't have to. Telepresence into robots for specific functions will undoubtedly be common, but for most purposes, like socializing, proxies win hands down.

Our current bodies are biological proxies. They are the only tools our brains have, at the moment, to interact with the world. Soon, we will enhance, and even replace them, with the help of technology. By replace, I don't necessarily mean remove our brains from them. We can liberate our brains from our flesh by rerouting our nervous system so that our brains can effectively inhabit other proxies, virtual and otherwise, by telepresence. And through those new forms, we will connect with other human beings like never before, in both quantity and quality.

People should stop saying, "I want my robot body!" and start saying "I want my virtual body!"

I try to envision virtual reality, virtualized environments, and proxies in Living Outside, so if you're interested in this stuff check it out. Episode 3 is particularly relevant to this discussion.

--
Brian

4 comments:

flamingsword said...

Robots also require parts, maintenance, upgrades, repairs, electricity, and storage space.

You're not figuring overhead costs to each scenario. Physical bodies require repair and maintenance as well, and rent on a brain vat assisted living community is still going to be a non-negligible cost. The best of all worlds is having a physical body while you're young, getting the implants, then retiring to a robot at a brain vat community. It should not be significantly more costly to warehouse a brain in a robot than a brain in a jar, and you're still mobile for emergencies.

Being rid of your failing physical body is going to be an important motivator, and as nothing else will be able to provide the emotional experience of picking up your grandchildren in a body that you identify with, robot has some edges you have not considered.

Brian Bazzell said...

Robot bodies with compact, mobile brain support systems will be considerably more expensive than brain vats. Brains vats have no space constraints and they have economy of scale with the ability to support a large number of brains with shared equipment. They will also have better internet connections than mobile robots and will not be as risky for your brain.

You don't have to put your brain inside a robot to identify with it. A robot designed for telepresence could give the same experience of being present at much lower cost. It will also have better performance as it would be unhindered by a brain support system. There could be an issue with lag over long distances depending on your activity, but you could move your brain closer to where you want to telepresence if that became a problem.

If your grandkids were in China and your brain was in Florida, you could move it to Germany and have effective telepresence in China and much of America, depending on how much of a problem lag will be. You might even move your brain to a new vat by robot, but keeping it in a robot day to day and moving it around to see people would be expensive and wasteful in terms of energy.

flamingsword said...

So let's do both. What about modular vats that can be loaded into robots for special errands, but are usually kept in brain support systems? You could have an operation like the underground robotic car park, only brains instead of cars.

Each person will probably have a favorite excursion robot, or maybe a private robot if they're very wealthy. Eventually they're going to learn the handling of that particular robot very well, and come to partly identify with it like people do with cars and computers. It will get co-opted into your sense of self through familiarity and repeated use.

And just because it's inefficient doesn't mean people won't do it. We do all kinds of inefficient things for emotional reasons. When your body is largely virtual, the thing you're going to start identifying more with your brain. You want your brain to be there for sentimental reasons that make no logical sense, but you still want them. The cost/benefit ratio is only going to be worth it for special occasions, but there will be some of those: weddings, children's birthdays, funerals, ceremonies. Don't just assume rational behavior, because people are not like that.

flamingsword said...
This comment has been removed by the author.