The Emperor's New AI

'It looks like you're trying to have a conversation with a computer. Can I help?'

Look! A talking dummy.

The pop media's fascination with "intelligent" computers, especially of the talking variety, shouldn't surprise us. It only mirrors our own anthropomorphic tendencies - to give things far more human characteristics than they really have. Whether it's voices coming out of the static, or faces in wardrobes or cheese toasties. The inanimate Golem brought to life, through either human or divine intervention (or both) is a myth that's taken many forms over the years.

As a result, AI has attracted far more than its fair share of flakes, phoneys and the outright naive over the years.

The cost is hard to calculate. There's an obvious resource issue, an opportunity cost, when fatuous endeavors are allowed to crowd out more pressing computing problems. Of all the woes we have with today's computer systems, their inability to hold a conversation must be one of the least important. We'd rather see systems that don't fail, that never lose data, and photographs that we know we'll be able to see in thirty years' time. Today's digital data is designed to be lost, it seems; imagine a generation with no family album, because the ink has bled and the formats can't be read. It isn't science fiction so much as a probable outcome.

And even if an "intelligent" computer was to be devised, it would help us a lot less than we might imagine. The world isn't short of intelligence. It's just very rarely applied to pressing problems.

So this kind of AI work is about as useful to us as research into how we can burn through our carbon fuels faster.

Fortunately, serious researchers may yet be able to shake off the curse of AI. In Manchester, where Turing made his flawed philosophical assumption that set academic AI haring down the wrong path for forty years.

At the University of Manchester Steve Furber, father of the ARM chip, is helping build a "brain box", modelled from biological systems. The research will help design more fault tolerant computing systems. There isn't a hint of talking (or dancing) robots in sight.

So to get better computers, maybe all we needed to forget about them being conscious, or intelligent. ®

Sponsored: Designing and building an open ITOA architecture