Machine learning is all the rage these days. Last year’s FT & McKinsey Business Book of the Year was “The Rise of the Robots”, a bleak prediction of how machine automation will take more and more jobs. In July a letter signed by 1,000 tech experts, among them Steven Hawking, Elon Musk and Steve Wozniak, warned of an impending “AI arms race” and in December Musk and the startup accelerator Y-Combinator announced OpenAI, an AI research institution meant to counter the threat of Robots taking over.
Apparently, if World War III ever comes around it will be between humanity and Google’s self driving cars.
These worries are not entirely unfounded. Especially the part where machines will take certain jobs (I sure hope you don’t drive a car for a living). But most of the recent progress that has been made has been in using neural networks and deep learning for pattern recognition. A machine might understand what you are saying, but not what you mean. There’s a lot more to human intelligence than pattern recognition.
In The Unreasonable Reputation of Neural Networks Luke Hewitt, a PhD student in in the MIT Department of Brain and Cognitive Sciences, writes
Deep learning has brought us one branch higher up the tree towards machine intelligence and a wealth of different fruit is now hanging within our grasp. While the ability [for machines] to learn … is both new and exciting, we should not fall into the trap of thinking that most of the problems an intelligent agent faces can be solved in this way.
Recent advances in neural networks mean that a lot of jobs are under threat. But they don’t mean that we are within reach of creating the next Skynet.
