Thursday, 8 June 2023

Leading the world?


A little over half a century ago, I sat in a seminar in which my then Psychology lecturer invited us to define ‘intelligence’. After we’d had a few tries, he put us out of our misery by telling us that the working definition is that ‘intelligence is what intelligence tests measure’. That is to say, people have developed tests to measure their concept of intelligence but there simply is no satisfactory all-encompassing definition of the word. Rationality – the ability to look at facts and data and reason a way to a conclusion is clearly part of it, but there is also such a thing as emotional intelligence, and it’s far from clear to me that that aspect is currently informing much of the thinking around Artificial Intelligence. Whether the dangers posed to mankind by AI are as serious as some are making them out to be is an open question. There are, however, plenty of scientists and experts in the field lining up to warn us of the dangers, and I can certainly empathise with the idea that any truly intelligent entity which looks at the current state of the earth is likely to conclude that the planet’s future might be better ensured if the plague of one particular species could be eliminated.

Let’s assume that the experts are right, and that AI poses a real and present danger to humanity. The proposed solution – greater regulation of those working in the field – seems to me highly unlikely to address the issue. If there’s one thing we know about the currently most intelligent species on this planet, it is that there will always be someone willing to break any rule that is made. There are, after all, laws against murder, but they don’t prevent murder, merely set out the process and punishment for handling the murderer after the event. Telling the world in a deep, profound, and multinational voice that they must not do certain things doesn’t really solve the AI problem, nor does having a process for punishing the transgressors after their products have destroyed humanity. In theory, the capacity of any computing processor to act should be limited by any parameters set by its programmers, but most people’s conception of true intelligence would obviously include an ability to consider the validity of those parameters and override them as necessary. Even Asimov’s famous laws of robotics don’t really seem to overcome the problem, because a truly intelligent machine would also necessarily have the capacity to challenge those. Perhaps it’s already too late: the attempt to put controls in place is an impossible quest. Or maybe we just need some AI help to solve it.

In the meantime, the UK’s Prime Minister, a man whose usual solution to all problems is to claim that they don’t exist and repeat his five doomed priorities, is busily presenting himself as the world leader on the matter. The basis for such a claim is dubious, to say the least. And whilst claiming to be world-leading in every field may play well to a home audience, I do rather wonder what impact it has on other world leaders when Sunak turns up at their meetings claiming to be setting the agenda and leading the rest of them. Probably not the impact he thinks he’s having.

1 comment:

dafis said...

Sunak is sticking his nose into the AI debate simply because his father in law is a major investor in a Big Tech company with a serious global reach. Follow the money is good advice in this case, no big deal or big social motive, just another plot/conspiracy to steer the "regulatory framework" in a direction where it looks good but has no balls.