One of Dr. Pearson’s main areas of interest is machine consciousness and the blurring of the lines between humans and machines (robots). Initially he feels that robots will help us in the home, particularly for the elderly. In the first instance this will be as sensors that detect changes in the person, alerting medical care if necessary. This will enable many older people to maintain their independence and stay out of care homes, which is not beneficial to them but also will help to keep the cost down.

Looking further forward though he thinks there will come a time when we do have actual automatons in the home. "I think robotics gives us an awful lot. Men and women die at different ages, and so this means that quite often women are spending a lot of years at the end of their life on their own. It won’t just be the physical assistance that robotics will provide, the artificial intelligence side of it is going to be very important too. The Japanese are plugged into that, they think it is very important that computers should have personalities, and be able to replicate cognitive conversational characteristics," he said. Dr. Pearson spent most of his time as an engineer thinking about how we might produce computers, which are very human-like in terms of their emotional characteristics.

As nano technology picks up and we get better techniques for picking up where signals happen in the brain, it won’t be very long before we can pick up signals from each individual neuron, which will provide insight into the way that the brain works.

"What I think is important is looking at how the brain achieves certain functions because we can then do derivatives from that, which might actually work better in many cases. It will certainly run faster inside a computer because signals travel much faster and we should be able to replicate conscious thinking with a full emotional repertoire in a machine."

Controversial

This opinion though is controversial and a lot of people in the field don’t think it’s very likely, but Dr. Pearson says he has never really understood why many believe that. To him there’s nothing magic in the brain, it's just an engineering system to be reverse engineered like any other and nothing that can’t be done with future technologies. He believes that it’s just a matter of figuring out what works and taking some of the concepts, building on those and doing it inside a computer.

As an example he thinks that with a 2D digital chip it would be very difficult to do artificial intelligence. As Dr. Pearson puts it: ‘a lot of stuff that happens in the brain is because the sensory system and the memory system and the processing system are all integrated into the same cell essentially. So that, to me, is an important part of consciousness, so if you have completely separate areas for memory, for sensing and other functions; I don’t think that works very well.

"I don’t think that there is that big a mystery to consciousness, I think that it’s something that we can learn how to do. We will be able to make computers within the next 15 − 20 years that are fully conscious and probably superhuman in terms of their capability,"he added.

Multi-core processors

What he sees happening is something along the lines of today’s multi-core processors. Back in 1990 he says he started thinking about this as he foresaw that Moore’s Law would end and the only way to progress would be with multi-core. What he sees, if you take the idea of multi-core as far as it will go, is a very large suspension of digital or analogue components in a gel.

Dr. Pearson believes that a computer in 2025 will resemble a pot of yoghurt where you have a suspension of processors in some sort of gel for providing them with power and some space interconnects such as lasers or LEDs to connect the computer’s processes up. Now when you do that it will have millions of connections between each of those devices rather than the few hundred on a conventional chip. So you get more connections and it's faster because it’s going at the speed of light through a gel rather than slowly through a wire.

Because things are a lot smaller the distances on the devices means that the electrons don’t have to travel very far as well. The speed of the device and the communication and the number of connections, all of these will have several more zeros than we have today. And if you add all of that together it’s possible to come up with a glass or pot of yoghurt that has roughly the same IQ as a human in terms of its number crunching ability. It’s a pretty scary prospect, he admits. The future CEO of a company might be a pot of yoghurt: yet it would be fully conscious, fully self aware but superhuman in terms of its capability.

Asimov

However; although he believes that machines will be like this he thinks that Isaac Asimov’s three laws of robotics are nonsense. "Asimov was a very good science fiction writer; but I think that's where he stops. He was a reasonably good scientist as well. I think he was wrong with his three laws of robotics. I think that any computer during its learning processes will learn how to read so that it can absorb human information and so it will know all about Isaac Asimov. It will know about these three laws and it will fully understand that discussion and the purpose of that discussion to hold itself in check. The computers that we are trying to keep under our control will fully understand that we are trying to do that to them and we can only do that with their cooperation.

"You can’t build three laws of robotics into the system because that's not how they would work. In fact we’ve already got robots out there in a few countries that are doing border monitoring functions and providing security services for big secure bases. Some of those robots don’t have three laws of robotics programmed in, they are quite happy to kill people, so we’ve already thrown Asimov’s laws of robotics in the bin.

These robots don’t necessarily think, but they do make decisions based on sensory input. So if they come to the conclusion that something is not a friendly person, that it is a threat, they are quite capable of deciding to shoot. They will do that; it’s just that at the moment in most military robots the decision to shoot is still taken by a human. It doesn’t have to be, it can easily be done by the robot."

Terminator

This of course is the ultimate nightmare scenario like the situation in the Terminator films and many people are going to struggle to cope. As he says, the science fiction angle, the dark side, isn't very far away, but if you think about that you can get very depressed. The other way you can think about it, as an engineer; is can it all be done safely?

"You can do it safely provided that you allow these computers to be connected in some way into the feeling of the nervous systems so that they have the same supercomputing ability that is also available to humans and then the computers don't get ahead."

He can foresee a problem though if the computers are allowed to get smarter than humans and there’s nothing we can do to keep up with them, then that is a security risk. If on the other hand some people have got a connection between their brain and the computer, which gives them the same sort of intelligence levels, then we still have people potentially in control.

Superhuman computer

Dr. Pearson thinks that the only safe way that you can progress down this route, and make a superhuman computer is when we develop a reasonable transparent link into the human brain so that things can keep up. He actually believes the future of human machine convergence is exactly that. To do this he thinks we will develop technologies such as active skin. This enables you to build chips into the skin’s surface, which then connect to your nervous system and pick up signals from each nerve ending in your fingers and record those sensations.

Gradually that technology gets better and you end up with implants in the brain. We will create sophisticated brain implants, whether for medical reasons or as upgrades to our brains and memory system and will eventually have IQ upgrades too. This will roll out over a period of time, but it won’t be long before we essentially have the ability to upgrade the human brain.

Smart people

He doesn’t think that the big question will be: how do we cope with smart machines, but: how do we cope with smart people? What he is talking about is upgrading humanity. He says that we can choose, to a larger degree, what we want to be like in that new world. The reason for this is because it’s not just IT that’s progressing but biotech with things such as synthetic biology. So in the same timeframe that we are developing superhuman artificial intelligence (AI) connected into our brain and IQ, other people are developing the genetic side of it and figuring out how to modify people’s bodies so that they don’t get certain diseases, how we can live longer and be more powerful.

Smart bacteria

Of course these technologies will converge and we will be building our computers into other organisms as well, even to the point where we have things like smart bacteria. This will be where we can modify the genome of a bacteria with simple circuitry in its own cell and at that point you get smart bacteria, which can then work together across the internet, effectively providing a means of scalable levels of intelligence. So you might have a pot of yoghurt but then you could also have lots of pots of yoghurt connected together as a super intelligence across the net.

Dr. Pearson readily admits that it will "get quite hairy" when we start mixing the biotech and the AI together but those are the characteristics of convergence. It takes nanotech, biotech, information tech and cognition and removing and enabling them to interact with each other.

Whether any of this will come true we will have to wait and see. However, considering Dr. Pearson’s past success it’s a good possibility.

Acknowledgement

This article was published in ITNow September 2009, magazine of the British Computer Society, www.bcs.org, and is republished with permission.

For more information please visit: www.btinternet.com/~ian.pearson