Voices of Science

Yes, well I’ve always had this dream, pipedream sort of wish list, because in my work it’s been conspicuous for ages that you get a lot of insight by just looking at potential vorticity distributions and the sort of dynamics that that informs you about is exactly the sort of dynamics that applies to the development of weather systems and so you’d think that it would help a weather forecaster to be able to see the PV distributions. So this made me think and I actually wrote a – an essay on this called 'A Dream of the Future' in which I imagined that the full potential of this idea was realised with, as it were, ace forecasters who’d become real experts at – at seeing the complicated three dimensional features of the potential vorticity and surface temperature fields. And aided by modern computerised visualisation perhaps some day in the future it would all be done holographically or something, somehow you could get a three dimensional picture in a very quick and intuitive way and perhaps there would be ace forecasters, part of who’s work would be to almost fly among these features and grasp their three dimensional nature and through their long experience and intuition see – spot the cases where the forecasts needed correcting. And this is rather against the usual culture in the weather forecasting world which quite reasonably says that, well no these things are too complicated for human perception to have any role at all and the whole thing has to be automated and computerised. 

What makes you think that it is a combination of the human and the computer that is needed rather than the computer?

Okay, that’s a very good question and I think it comes from recognising that human perception and intelligence is still much more powerful than our electronic computers, in some ways, and a good way to make that point is to point out that computers don’t yet drive taxis do they? And you can, you know, if you think about how perception works, you can I think have some insight into why: it’s because it requires massively parallel computing. You have to deal with a combinatorially large number of possibilities at once, that’s all related to this unconscious power of abstraction that we have. And electronic computers aren’t yet as good as we are at some of those things, so I think it’s an interesting open question as to what – what combination of human and machine intelligence is most powerful. I would argue that some combination is powerful and if we do things that use our powers of visual perception, which is a very sophisticated part of our perceptual apparatus, then there’s scope for doing better than we could do either by ourselves or the machines by themselves. So I still have that dream although it – you know, hardly got going as yet.