Mar 29 2007

Digital Immortality

Many science and science fiction writers have explored the idea of uploading the information in our brains into a computer substrate – to make a digital copy of ourselves, or to actually transfer our consciousness into a machine. The idea is interesting, but the looming existential question that is never fully answered is whether or not the computer version of you would really be you – often referred to as the question of continuity. This is not a trivial question – without continuity, you die and a mental doppelganger takes your place. With continuity, this can serve as a pathway to virtual immortality. I thought I would add my neurologist’s take on this question.

First, let me dispense with the easier (but not easy) question of whether or not computer-based intelligence can be truly self-aware. Books have been written about this, but I will just summarize the bottom line as this: there is no reason in science why a machine or computer based intelligence cannot have the same property of consciousness as does a human brain. However, information is not enough – the computer brain must have the structure and function to process the information so as to generate the emergent phenomenon of consciousness.

But let us assume for further discussion that sometime in the future we will be able to create an artificial brain – capable of true self awareness. Would it be possible to transfer one’s consciousness into such a vessel?

This is a vexing question. I take the strong position that we are our brains – and not just the information in them, but the hardwiring and processes of the brain. So simply copying the information from someone’s brain into an artificially intelligent (AI) receptacle won’t work. The AI may be truly self-aware, and they may think that they are you, but will just be a copy.

Some have proposed using a gradual process of replacing the biological brain with AI components, so that the process will be gradual and seamless, and eventually your brain will be replaced by an AI computer, but will still be you. This is an interesting proposition, but it is still possible (depending on the details of this is carried out) that you will still just be replaced by an AI copy, just gradually. The fact that your biological brain is destroyed in the process makes it worse – because not only is the AI not you, you are dead.

The critical component to being assured that any transfer does not end with the death of you is continuity – your consciousness has to continually exist as it is moved (not just copied) into the AI computer. So here is my proposal for how to do this.

For background it is helpful to understand that our brains, in a way, are actually two brains: the right and left hemispheres. Each hemisphere is capable without the other of being fully awake and self-aware (although each only contains a piece of our neurological function). In fact you can pharmacologically “turn off” one hemisphere or the other and the subject will remain awake and aware, they will just lose the specific functions that exist only in the other hemisphere.

The two hemispheres of the brain are connected by several large bundles of axons – the largest being the corpus callosum. These connections provide massive real time communication between the two hemispheres, and the result is one seamless mind or consciousness. So we are not aware at all that we have two minds – they behave as one and for all practical purposes they are one.

Now let’s imagine that we develop an AI brain that was capable of reproducing the processing of the human brain, but was much more powerful and contained a much greater memory capacity and fidelity. Further, this AI brain was capable of communicating to and receiving input from a biological brain. Therefore we could make massive connections between the AI brain and a human brain. These two brains would then have a similar relationship to each other as the two hemispheres of the brain do.

The AI brain could be designed to model itself after the pattern of processing that it reads from the biological brain – so that eventually it will completely redundantly reproduce each function within the biological brain. If it does this precisely enough (which is feasible even with conceptually simply mapping algorithms), the AI brain should be able to reproduce the personality and feelings of the biological brain. It will also increasingly over time redundantly map the memories of the brain.

At first the recipient of an AI brain augmentation might not notice anything dramatic. Perhaps they would notice that their ability to learn and remember new things is vastly improved. As they begin to think more and more with the combination of both brains working as one, they will notice that their thoughts are quicker, and that they have augmented mental abilities. The AI brain could be pre-programmed to be very good at math – so over time you would evolve into a version of yourself that has a great memory and is phenomenal at math. In fact, all specific cognitive functions can be enhanced by the AI brain.

Most importantly, throughout this entire process your brain continues to function and be the substrate of your consciousness, but is has just extended into and expanded (AI) anatomy of memory and processing. During this phase of your existence you will be a hybrid AI/biological brain. Because this is an evolutionary and non-destructive process, perfect continuity is maintained. After a while, the AI portion of your brain will be as much you as the biological portion – just like the right half of your brain is as much you as the left half.

It is possible that during this period when your biological brain goes to sleep you will experience awareness solely with the AI portion of your hybrid brain. Similarly it may be possible to turn off your AI brain and experience pure biological consciousness once again. You will still be you – but you will lose vast portions of your memory and abilities. This is exactly analogous to what happens when we put one hemisphere to sleep.

Eventually, after sufficient modeling has taken place and your processing has merged sufficiently with the AI brain, you may not even notice when your biological brain goes to sleep. The AI brain may have hundreds or thousands of times the memory and processing power of your meat brain, so the loss of your original brain may represent a loss of only about .1% of your neurological function – negligible.

Eventually your biological brain will likely die, and you will hardly notice it. Perhaps before that happens you will find it convenient to have your AI brain transferred into another body, which can be either biological (even a clone of your original body, you know, for sentimental reasons) or a robot or cyborg – whatever you choose. The transformation to digital immortality will be complete.

There are other interesting consequences to such technology also. Once you are predominantly an AI brain, that brain can easily be connected to other computers, even wirelessly. You could enter a virtual universe that would seem 100% real to your AI brain (of course that has some scary implications too.) In fact the programming for the virtual universe could be built into your AI brain – so that you could enter your internal universe whenever you want, and even invite others in to play.

Also, AI brains could be fixed with wireless transmitters and receivers. Telepathy would finally be a reality – in fact it would likely be a much more efficient form of communication. Communicating by flapping meat together and vibrating tissue will seem quaint and outmoded. We could also communicate this way with our technology – we won’t just drive a car, we will become the car. Writing and creating art will be a matter of simply imaging the end result and having it appear.

How far away is all this? Ray Kurzweil and other thoughtful futurists think it’s about 40-50 years off (see The Singularity by Kurzweil for a thorough discussion). It certainly seems highly probable that we will have the raw computing power by then. As I wrote about before, we are already working on developing the brain-computer interface. It’s of course difficult to predict the future, which is likely to have both unexpected obstacles and unforeseen new technology. But I think that something like this is not too far off, either in the latter half of this century or soon in the next. Whatever happens it’s likely to be more fantastic than has yet been imagined in science fiction.

No responses yet