HOW CLEVER ARE TODAY'S COMPUTERS?

Written by David Tebbutt, Mensa 03/89 - scanned

Hello. Welcome to my first bi-monthly contribution to the MENSA MAGAZINE. Since I'm in the computer business, you won't be surprised to learn that these columns will be about computing and related issues. (No. Don't go away. I promise to write in plain English and to avoid boring, incomprehensible, technicalities.)

Have you noticed how all-pervasive computers are becoming? Shop tills are connected to them, high streets seem to sprout new money in-the wall machines weekly, no respectable office is without at least one personal computer and factories are turning more and more to automation in their struggle to stay competitive.

You would be forgiven for wondering where these soulless artefacts leave you, the human being. If you're part of the computer in-crowd, then computers probably don't bother you too much. But if you're an outsider you probably feel at least perplexed, if not downright threatened, as you perceive computers 'taking over' in all walks of life.

The computer seers don't help much, with their claims that one day the computers will do everything for us. The theory runs that the computers will create the nation's wealth and we'll have a life of ease and luxury. This pre-supposes that the wealth is distributed in an equitable fashion and it doesn't just pour into the hands of the computer owners. Quite what we'd do with all the leisure and no work to give our lives structure and meaning, I'm not sure. And, some of us (the lucky ones?) will still have to work anyway. I have real difficulty imagining a computer writing love song or a powerful novel, for example, and I'm not too keen on having one performing open heart surgery or giving a blanket bath.

But, whether we want it or not, researchers around the world are working on ways of making computers think as well as, if not better than, us. The advantages will derive from the computer's access to almost unlimited information and its ability to manipulate, filter and do calculations on it at very high speeds. The end result is clearly far ahead of anything an individual could hope to achieve. Hence the widespread feeling that computers are somehow 'cleverer' than us. They are but, for now, only in strictly narrow domains.

I liken the computers researchers' quest to an earlier race to split the atom. The scientists were driven by their thirst for knowledge. They knew their achievements could be turned to good or evil ends but they hoped common sense would prevail. Of course, they knew little of the side effects of massive doses of radiation or that the world population of nuclear reactors would lead to such horrendous long-term waste disposal problems. They must have suspected the bomb-making potential but they could not possibly have imagined the kind of world their technology has led us to.

And it's not just the atom-splitters. Chemical engineers have created deadly substances which are corroding the atmosphere. Biotechnicians have created some awful bugs which will hopefully never find their way out of a germ warfare canister. My guess is that most of the scientists behind these developments thought only of the potential benefits of their work. Sadly, real life just isn't like that. And I don't see why the results of artificial intelligence research should be treated any differently. Once the genie's out of the bottle it will be awfully hard to stuff it back in.

The good news, at present, is that most computers aren't anywhere near as clever as they're made out to be. If you've got a personal computer you may find it responds to the word 'type'. It isn't usually programmed to cope with 'tpye' or 'tyep'. But we know immediately what the computer user is getting at.

Once you start to get a computer to recognise mis-typing, you need to define where its intelligence will end. The computer doesn't usually try to be clever unless it's asked to. Many people use spelling checks to run through documents after they have been prepared. The spelling checker will spot all misspellings, except when they result in valid words. To overcome that problem you'd need a computer that understood grammar, that could divine context and know what words were likely to be used in that context. Suddenly, you're talking about problems which might hog the computer's power without delivering a significant enough improvement in its results. So, right now, humans are both smarter and cheaper for the sort of task.

I think we humans have still got a lot going for us. To take a topical example, look at desktop publishing. Is there anyone who hasn't been exposed to the hype for this latest computer cure-all? The theory is that you buy a desktop publishing system and, suddenly, you life is transformed. Presentations, reports, books, overhead projection slides, newsletters, paper and magazines will suddenly start pouring from your computer.

Never mind the fact that you couldn't design a page layout to save your life. Never mind that it takes weeks to learn how to use the system properly. The truth is that like any tool of the trade (a dentist's drill for example), the computer can only produce quality results in the hands of a skilled user.

Sure, desktop publishing systems can (for a significant investment of time) produce better results than the word processor or typewriter. But, for top notch results, you need professional designers. They can draw on a vast body of knowledge and experience and apply it to the needs of the job in hand. Before designing a brochure, for example, you need to understand the target audience. You use this knowledge to create a brochure which will appeal to that type of reader.

The artificial intelligence brigade will tell you that all this expert knowledge can be captured and codified and stuck inside a computer. I've been hearing this for years now, and I'm still not convinced. But let's assume that human creativity can be fixed in silicon. Do we really want the sort of world this would lead to?

I say no. How about you?