Ask Glenn

Will computers ever be cleverer than people?

Difficult to say, but it probably depends on what you mean by ‘clever’. Many computers today can already ‘think’ faster and better than people, but only in very limited ways. For now, at least, computers are not very intelligent or clever by our standards. In the future – who knows?

Robot doing Rubik Cube

But I thought some computers could beat people at chess and stuff. Doesn’t that mean they’re pretty clever already?

That’s true, a particularly powerful computer called Deep Blue beat the world champion Garry Kasparov in a chess match in May 1997. But that only worked because winning at chess is ultimately based on how many moves ahead you can ‘see’ (or predict the results of ) and using this to decide the best moves. Using raw computer power, Deep Blue could predict possible moves up to twelve moves ahead. Even the best chess players, like Kasparov, can only predict up to about ten moves ahead.

So the computer was cleverer, then?

Not really. All that game proved was that a computer could store and handle more bits of information about chess moves than a human brain can. But being good at a board game isn’t a real measure of ‘intelligence’. Even if it were, there are other games that computers are pretty hopeless at. For example, in the more complex Japanese board game ‘Go’, even the most advanced computers can’t even beat novice Go players. The same goes for complex card games like Poker, which computers are no good at because they can’t bluff (or even cheat) the way human players do.

Why not?

Because with games like these, using calculations alone won’t work. To play good Poker or Go, you need other things like intuition, creativity and even empathy – which is the ability to imagine how someone is feeling. Simple as they are, games like these help to demonstrate why information processing is not the same as intelligence, and why even the most powerful calculators can’t be called ‘clever’. At least not yet.

So what makes us so much better than them?

In short, our brains. The average human brain weighs about 1.5kg – about the same as a bunch of bananas! – but contains about 100 billion neurons, or nerve cells. In computer terms, it uses these to store about 100 million MB (megabytes) of information, which it can handle at a speed of over 100 million MIPS (million computer instructions per second). While some of the most advanced computers around today may just be able to top this, it’s not speed or memory power, but complexity and flexibility that gives us the edge.

We’re not entirely sure how it all works yet, but we know that because each neuron in the brain connects to an average of 1,000 others, this creates over 1 quadrillion connections. That’s a big number. If you tried to write it, it’d be a one with twenty-four zeros after it! It’s this vast number of connections that makes the brain so powerful. It allows us to go beyond simple calculations and into the realms of emotion and consciousness – or thinking for ourselves. Until computers can be made truly conscious, we won’t really have created true artificial intelligence, or AI. Judging by what we have now, we’ve still got a long way to go.

How will we know when we’ve done it?

Well, for one thing, it’ll pass the basic test that all computers so far have failed.

What test?

The Turing Test. This was created in 1950 by Alan Turing, one of the scientists involved in building the first modern computers. He said that the real test of artificial intelligence was this: if a person talked (through a keyboard and computer screen) with another person and a computer and couldn’t tell which was which . . . then you’d have created a truly intelligent computer. As of 2007, no computer has ever passed this test.

That what you think.

Eh?

402*Error* insufficient memory/ 640K barrier exceeded%$^

Ahhh!!!! All this time I’ve been talking to a computer?!

Ha, ha. Gotcha.

Grrrrr.