It’s Alive. Aliiiive!

Twenty years is a good time period in futurism. It’s close enough for people to get excited, and for many to think that they ought to see the day. At the same time, it’s far enough away that if you’re wrong, memory of your prediction will probably have faded. Even if it hasn’t, it’s old enough news that no-one really minds. Keep an eye out for it; if an advance is predicted to come about in twenty years, it usually means that no-one has any idea how long it will take. The other biggie, of course, is ‘in the next century’, which means either “it’s coming, but we don’t want to scare you,” or “it’s so implausible there’s no chance you’ll see it, but we don’t want to annoy you.”

One of the advances that’s been most persistently twenty years away for decades and decades now is artificial intelligence. There have been a few modestly encouraging achievements — neural networks, expert systems, and so on — but in reality, software experts have been totally unable to even begin trying to program an intelligent piece of software. Rumours abound regarding the sentience of Echelon II/Magistrand, the English-speaking world’s massive spy computer — the one that scans every electronic communication — but they’re just not plausible. It would be easier to believe that a sentient Magistrand was a gift from aliens than that we’d programmed a real AI.

The main trouble is that the human brain is just too complex. We don’t really know how it works yet, and there’s even uncertainty regarding the function of different areas. If we can’t even understand consciousness, how can we hope to program it?

Henry Markram: Would you buy a shark-mounted laser doomsday device from this man?

Professor Henry Markram is a South-African Israeli genius, a doctor turned computer engineer, has announced that he expects to have a working electronic simulation of a human brain by 2018. Just eight years. His approach is refreshingly direct. Rather than try to work out how each different brain function works and program flexible simulations, he plans to use dissection and examination. His goal is to map all of the physical elements that make up a human brain, and then reproduce a duplicate structure inside a computer.

The work is intricate and painstaking, but none the less achievable. He has already reproduced — simulated? uploaded? — a mouse’s neocortical column. It’s just computer power stopping him from producing an entire mouse-brain, or starting work on recreating a human brain. Markram is based in Switzerland, and is getting huge amounts of funding for his Blue Brain. As the name suggests, IBM are providing a lot of help and support, too. Given the ever-increasing power of the computing industry, he expects to have sufficient computer power in the next few years.

The result, of course, remains unpredictable. We don’t even really know how memory is encoded, so it’s difficult to guess how the Blue Brain will manifest. But his simulated chunk of rat-brain has occasionally produced totally unexpected bursts of brainwaves over the last year. Some scientists have predicted that the Blue Brain will be an ‘empty bucket’, nothing more than a clever model; others have pointed out that humans start out as fairly blank slates too. Regardless, Markram relishes the issues that his work will raise:

“The process of building this is going to change society. We will have ethical problems that are unimaginable to us”