It seems to be a given by most scientists these days that we are on the verge (within 25 or 50 years) of Artificial Intelligence. Like gravity, the closer we actually get to this, the more people will become involved, wishing to be the first, the more likely it will be to happen. I think it could be described as a probable event now.
We are also seeing a lot of articles warning us of the technology and moral implications. The Guardian (http://www.theguardian.com/science/audio/2014/aug/04/science-weekly-podcast-nick-bostrom-ai-artificial-intelligence) and some of our top thinkers have weighed in too (http://www.cnbc.com/id/101774267#.)
So, the paradox is, it is getting closer but no-one knows the possible results of it. We could be sowing the seeds of our own destruction without knowing it.
I am going to make a suggestion. Let us not model A.I. on the human brain, but link it with the human brain. In informational and emotional terms. Modelling A.I after the human brain is observing the physical qualities of how our brain works. But these qualities are created by a combination of d.n.a. and the information we fill that brain with. Both of these are simply numbers, codes. We have one of the codes already and we are developing the other each year with more and more accuracy.
1. Human dna contains the code to build a human body and brain. We have mapped our genome, but we don't understand how it all links together yet. But it is another part of science that seems to be a probable event. It merely takes time, computers, experiment and observation.
2. We are modelling virtual reality which is becoming more and more sophisticated. It might be a long way off to fully recreate a virtual earth, but one just has to play a modern console game to see how close visually we are getting. At some stage in the future even we will not be able to tell VR reality from 'real' reality. It is not there yet, but as a prediction I don't think it is unreasonable to assume we will develop a convincing 'matrix' style VR world within 25-50 years
3. Add these two together. Why? We need to model the human brain not on one brain, but on the cumulative processing power of all of us. We need a super intelligence, but one that is connected to us, one that feels the same way as we do. Without our weaknesses and our failings, but one with our desire and capacity to love. To make the A.I. not just human, but super-human. We can't simply input dna and hope for the best. We need some sort of selection process that allows us and the super-intelligence to choose dna that will be the most balanced between love, empathy, intelligence and rationality. A VR world is the most logical solution. And dna strands simply choose themselves to be added to the super brain.
Our aim is to make a virtual human brain by slicing together dna codes from all of us -which is more or less what nature does anyway - to get the most efficient method of recreating our existing thinking organ. VR simply allows the machine to remodel reality to get the best results.
Each dna code is a 2d virtual brain, but a human one. As we add a new 'slice' it is still 2d, but layer upon layer of 2d suddenly gives the virtual brain a further dimension. It makes it 3d now. The superbrain exists in our world, the 2d virtual brain exists in the virtual world.
This is, as far as I know, the only way we can create a super intelligence, but also know that it is also the same as us, it is not alien to us and hopefully it will not try to kill or control us. Any other descriptions or attempts to create superintelligence are all doomed to a 'try it and see' approach, but this approach ensures a more likely positive future. Why? Because the superbrain is simply the combination of all of us. It is part of us and we will be part of it.
Sounds a bit sci fi again. But so does an Artificial Intelligence. We are not smart enough to create an artificial intelligence, so we should concentrate on creating virtual human intelligence and allow it to create itself as it were.
Of course, if this happens, no-one will ever be sure they are in the VR world or the 'real' world. Or even both if the technology is there. Can we recreate lower resolution versions of ourselves? If the answer is yes, the the next question is: can we create higher resolution versions of ourselves?
So, here is my suggestion: simply design computers to help us model virtual reality and the human brain. Sooner or later no matter how unintelligent the programme that runs these models are, the code contained in our dna will at some stage achieve self awareness. We did. We don't need to design an A.I. we simply have to model dna correctly and put it into a virtual world that would be recognisable to ourselves. Sooner or later the A.I. will actually design itself.
(As a postscript, as far as the virtual 'us' are concerned, they live and die and are joined to the force that created them - the superbrain that lives in our world. In effect God for them. A bit spooky when you think about it)
No comments:
Post a Comment