Computers as confidantes

Written by David Tebbutt, MacUser 06/90 item 03 - scanned

IBM's press releases are usually of two types - boring and very boring. Imagine my surprise when one actually grabbed my attention and held it. The chilling title was 'IBM super computer based model produces waves similar to those in the brain'. The press release goes on to say 'a computer model for studying the brain has unexpectedly produced - on its own - electrical waves like those actually found in the brain itself.'

Apparently, an IBM scientist and a couple of Columbia University researchers were modelling the behaviour of 10,000 brain cells in the hippocampus, a part of the brain which is essential to the formation of new memories and quite frequently the origin of epileptic seizures. (It also is the Latin name for seahorse. Bet you didn't know that.) The researchers hoped to gain a better understanding of epilepsy while IBM thought the research might suggest new ways to design computers.

Initially, the team used the model to confirm things they saw in the laboratory. Now they experiment on the computer model as if it were an organism in its own right. In the 10 years since the study began, the researchers have gained what IBM describes as 'valuable and unexpected insights' into epilepsy. Now that these unexpected waves have been produced, the researchers are even more convinced that their model of the hippocampus is an accurate one. No-one knows why the brain produces the waves, so the team will now use the model to try and unearth their cause and function. It's eerie isn't it?

In the future, the group expects to be able to investigate how people think, learn and remember. A mammal's brain contains billions of neurons, each connected to hundreds or even thousands of others. The connections are so complex that no-one has yet figured out what the organising principle is. The IBM scientist, Roger Traub, says, "If there is a structure to the way all these cells are connected, we don't have the tools to see it."

Still, it is interesting that computers are beginning to display the characteristics of living organisms. How long will it be, I wonder, before the computer itself helps scientists discover the underlying principles of how our brains work? Then it will be only a matter of time before this is converted into computer programs and we will start trying to fashion organic intelligence from silicon. IBM, as usual, will be at the front of the pack. With its massive profits, it can fund huge research projects, even in these recessionary times.

Will we end up with computers that can truly think? To me this has always been laughable, but now I'm not so sure. Are the big-technicians in an evil race against the computer researchers to be the first to build alternative life forms? I have to confess that I rather like being part of Homo Sapiens, the dominant species. I don't fancy being superseded by either homunculi or computers. With their lightning-fast brains, infallible memories and instant access to vast remote databases, I'd be inclined to place my bets on the computers to win this particular race.

I wonder what sort of machines we'll be sitting in front of in 50 years' time? Will we even recognise them as machines or will we simply regard them as friends, colleagues or bosses? Psychologist Neil Frude has long suggested that once the underpinning technological issues have been resolved, attention will turn to how the machines can be best presented. He talks of computers which can look like fluffy cats being supplied to old ladies. The cats will feel warm to touch, will purr when stroked and will remind their owners when they're due to take their pills.

The machines of the future could theoretically get to know their owners and begin to anticipate their needs. Perhaps scanning newspapers or incoming data streams for information likely to be of interest. Or maybe endlessly listening to the same old stories and always giving fresh and lively responses. We already know that people are willing to confide in computers and that they quickly forget they are dealing with machines. Joseph Weizenbaum, to his eternal regret, once wrote a program called ELIZA which simulated the behaviour of a Rogerian psychologist. It would display a message like, "What's wrong?" You might reply "It's my Mac". ELIZA would reply, "Tell me more about your Mac." And so it would go on, picking up key words by parsing your inputs. But many people believed that ELIZA was truly intelligent and could help them.

The computers of the future will be far more convincing than ELIZA and I wouldn't be at all surprised if we don't end up forming more satisfactory bonds and relationships with them than those we presently have with human beings.