← Archive
Vol 70 No 23 7 February, 2019

AI

By Stephen Bredenkamp

AI

The Turing test, developed by Alan Turing in 1950, is a test of a machine’s ability to exhibit intelligent behaviour equivalent to, or indistinguishable from, that of a human - Wikipedia For many years, gaming was thought to be the pinnacle of AI. Beat a human in chess, or similar ‘complex’ game, and you show ‘human intelligence’. However, it was eventually realised that all that it took was a very quick, very big computer and it was feasible. The computer could effectively analyse an ‘infinity’ of possible variations. No intelligence required – just brute strength. The development of AI has now reached a new level, with ‘deep learning’, or ‘machine learning’ whereby a computer can ‘teach itself’. With the latest developments, the rules of chess (or Go, or …) were fed in and the computer ‘taught itself’ how to win. Within 4 hours it could beat any other ‘brute strength’ based chess computer. So could we now say the computer is ‘intelligent’? This question is still prominent in philosophical and neuro-scientific circles, with but one of the accepted answers being that we have infinite possibilities with ‘Artificial Reasoning’, which is very deterministic, and not yet intelligence. Intelligence, in this sense is undefined – even intuition, gut-feel, etc has been argued away as the unconscious consideration of past history, developing a conclusion you are not even aware of. What is without doubt, however, is the change this will make to society. Even if machines are never ‘intelligent’, ‘artificial reasoning’ – the creation of an answer to any deterministic problem is with us. Autonomous cars (no driver needed), any assembly line, office job, in fact any activity that can be ‘deterministic’ (i.e. is rule based) can be performed by a computer program. It is purely a technical challenge. China has just landed a rocket on the far-side of the moon! This long introduction was my response to an article on the BBC web, ‘The last things that make us uniquely human’. It has profound implications for society, and particularly education. What indeed is the need for learning a process, or a fact, if a computer can do it infinitely more accurately, or reveal it within seconds? The article goes on:Actually, it all comes down to a fairly simple question: What’s so special about us, and what’s our lasting value? It can’t be skills like arithmetic or typing, which machines already excel in. Nor can it be rationality, because with all our biases and emotions we humans are lacking.