“The ever accelerating progress of technology … gives the appearance of approaching some essential singularity in the history of the race beyond which human affairs, as we know them, could not continue.” – John Von Neumann
In 1997, Deep Blue beat Garry Kasparov at chess by calculating using brute force the possible next moves and choosing the best one. No human will ever be able to beat A.I. at chess.
In March 2016, Google’s AlphaGo software beat Lee Sedol at Go, using a new method called “deep learning”. This new technique enabled a computer to learn on its own, playing against itself to improve its game.
One of the few games humans are still undefeated by machines at is crosswords, because machines still find it difficult to infer that a three letter whiskered animal that hunts mice and loves milk is a cat. However even here AI promises to beat us in the near future. Currently Google Translate’s AI is capable, after being taught to translate from English to Japanese, and from English to Korean, to translate directly from Japanese to Korean, without using English. This means the AI has created its own language!
Moore’s Law observes that computing capacity doubles every year. Seeing where we are today and projecting this accelerating rate of advance into the future, it is probable that at some point we will be able to create an AI which is as smart as a squirrel. Some time after that as smart as a chimpanzee. At some point AI will reach human intelligence, but there is no constraint stopping it there. Once AI is capable of enhancing itself past the level of human intelligence, very fast we humans will be closer to the intelligence of a squirrel than to the AI.
The implications of this computer superintelligence are equally fascinating and concerning:
- How can we ensure this AI has human values? In order to fulfil any task the AI may set out to solve, human interests may be in the way. When asked about the implications of Artificial Intelligence, Stephen Hawking said: “You’re probably not an evil ant-hater who steps on ants out of malice, but if you’re in charge of a hydroelectric green energy project and there’s an anthill in the region to be flooded, too bad for the ants. Let’s not place humanity in the position of those ants”.
- If AI is capable of running the world, performing every single job. What will humans do? How will income generation and inequality look like in a world with no jobs? Bill Gates proposes that companies which use robots pay taxes for them. Elon Musk points towards a basic income which governments would have to pay citizens .
- If this AI goes astray, can we turn it off? Probably we will have become so dependant on it that turning it off would not be simple. Where is today the off switch to the internet?
- How will humans interact with this AI? Will a human marry an AI? Will we have AI CEOs or presidents? Will this AI colonize and terraform other planets for us?
The same way the atom has always been there waiting for humans to discover it, it’s more a matter of when than a matter of whether humankind will be faced with this problem. If we get this right and we manage to survive it, future generations will probably look back and agree this was the single most important thing humans have achieved.