Monday, 8 March 2010

Donald Mitchie: on Machine Intelligence, Biology & More – Ashwin Srinivasan (ed.) ****

This is an eclectic collection of writings by and about Donald Michie, the Scottish-born scientist whose career spanned over half a century and covered many topics, most notably computer science and reproductive biology. Michie died in a car accident in 2007, aged 84, and “Machine Intelligence” is a tribute to his life and work compiled by the eminent computer scientist Ashwin Srinivasan.
The book varies widely in style and subject matter, but it is interesting and readable throughout. It comes in three parts, “Machine Intelligence,” “Biology,” and “Science and Society.” Each section is divided into chapters containing 3-5 pieces, with helpful introductions to the chapters by Srinivasan.
The writing is aimed at the non-specialist reader, and specialists may be disappointed by the absence of any of Michie’s many ground-breaking scientific papers. The upside is that experts and novices alike are treated to insider accounts of Michie’s code-breaking at Bletchley Park during WWII, reflections by Michie on how scientists work and the role of government in science, and thoughtful discussions of big topics in AI – such as the Turing test and the role of subconscious or “inarticulate” thought in cognition. Especially worthwhile are Michie’s thoughts on the difference between brute-force solutions to computing problems and truly intelligent solutions.
Michie was much more than a scientist, and some of the most witty and enjoyable writing in the book sees Michie as science administrator, social commentator, and popular science writer. Some of my favourites are his cutting comments on the Lighthill Report (the government report in the early 1970s that almost killed Britain’s nascent AI industry), his article about the reading habits of scientists (they do surprisingly little), and his account of a bizarre trek from London to Moscow that Michie undertook at the height of the Cold War.
“Machine Intelligence” is not a detailed or systematic treatment of Michie’s ideas – it’s a series of snapshots rather than a portrait. Articles on the same theme (like the difference between clever and intelligent computers) are sometimes scattered through the book rather than grouped together. And there are too many typographical errors. But “Machine Intelligence” succeeds as a readable tribute to a remarkable man, giving many glimpses of Michie’s insight, humour, and wide-ranging enthusiasm for science.
Review by Michael Bycroft

No comments:

Post a Comment