Professor Donald Michie

Donald Michie, computer scientist: born Rangoon, Burma 11 November 1923; Research Associate, London University 1952-58; Senior Lecturer in Surgical Science, Edinburgh University, 1958-62, Reader 1962-67, Director of the Experimental Programming Unit 1965, Chairman, Department of Machine Intelligence and Perception 1966, Professor of Machine Intelligence 1967-84 (Emeritus), Director, Machine Intelligence Research Unit 1974-84; Professor of Computer Science, Strathclyde University 1984-92; Director of Research, Turing Institute 1984-86, Chief Scientist 1986-92, Senior Fellow 1992-94; married 1949 Zena Davies (one son; marriage dissolved), 1952 The Hon Anne McLaren (DBE 1993, died 2007; one son, two daughters; marriage dissolved 1959), 1971 Jean Hayes (née Crouch, died 2002); died 7 July 2007.

In the 1960s, Donald Michie founded the field of machine intelligence in Britain through a unique combination of personal history, political savvy and academic brilliance. He had become attracted to the field of machine intelligence during the Second World War, when he had come to know Alan Turing - the most influential computer scientist of his generation.

Although there had been a "cybernetics" movement with an interest in intelligent mechanisms since the late 1930s, it languished after the war and Michie was unable to find a place within it. He therefore set out on an academic career in genetics. Michie had been convinced by Turing that the key to machine intelligence would be the availability of powerful digital computers. These started to arrive in British universities in the early 1960s and Michie adroitly switched disciplines, eventually establishing the Department of Machine Intelligence and Perception at Edinburgh University.

Donald Michie was born in 1923 in Rangoon, Burma, the son of James Kilgour Michie and his wife Marjorie. He was educated at Rugby School and in 1942 won an open Classics scholarship to Balliol College, Oxford. At Oxford he had planned to study Japanese, which he thought would lead to interesting war assignments, but because the course was full, he opted for instruction in cryptography instead. He turned out to have a great facility for code-breaking, and in summer 1942 was assigned to Bletchley Park, the wartime code-breaking centre in Buckinghamshire.

His most important contribution to the war effort came in spring 1944. At that time an electronic code-breaking computer, the Colossus, had been operating for some months. Michie and a colleague, Jack Good, made improvements to the machine which led to a several-fold increase in speed. Michie later recalled that "at the end of hostilities, nine new-design Colossi were operational and 63 million characters of high-grade German messages had been decrypted". Before he was demobilised, Michie was given the task of writing a detailed technical history of the code-breaking activity, which was only released into the National Archives in autumn 2000 - when his extraordinary technical achievement became publicly known at last.

During their time at Bletchley Park, Michie and Good enjoyed Turing's company, and they discussed Turing's idea for a "child machine". This machine - like a human being - would have an innate propensity for intelligence, but would have to learn about the world in order to exhibit intelligent behaviour. Michie was gripped by Turing's conception, but also realised that computers would have to grow up before it was possible to take the idea forward. In the meantime, he decided to become a geneticist.

He returned to Oxford in 1946, obtaining an MA in human anatomy and physiology in 1949 and a doctorate in mammalian genetics in 1953. In 1952 he had married Anne McLaren, a fellow genetics research student (an earlier marriage in 1949 had been dissolved). They both took up research positions at London University, where they made genetic studies of mice and produced several joint publications.

In 1958 Michie took up a senior lectureship in the Department of Surgical Science at Edinburgh University. McLaren also obtained an appointment at Edinburgh, but they divorced in 1959. Nevertheless they remained close, life-long friends.

Michie had retained a hobbyist interest in intelligent machines and, goaded by a sceptical colleague, designed a program that would learn the game of noughts-and-crosses. Lacking access to a real computer, he simulated the program's behaviour by a device consisting of matchboxes and glass beads. The contraption proved his point, however, and this led to his being invited to program the system for real at Stanford University, in California, which owned a large IBM computer.

Returning to Britain, Michie was depressed by the lack of comparable computing facilities. He discovered he had an aptitude for lobbying, and through personal appeals, lectures and newspaper articles, he argued for the provision of better computing facilities in British universities. His message began to get across and he earned a reputation as an important academic voice. Michie was promoted to a readership at Edinburgh in 1962, but dissatisfaction with surgical science and his growing confidence in the computer world induced him to break away to form an Experimental Programming Unit - essentially as a skunk works. Michie was good at writing research proposals and, no doubt mollified by the research money he was bringing in, the unit was given official recognition by the university in 1965.

In 1965 Michie organised the first Machine Intelligence Workshop in Edinburgh. Together with the published proceedings Machine Intelligence, the biannual workshops became the major forum for British researchers. Michie attracted two heavyweight academics to the unit, the psychologist R.L. Gregory and the distinguished theoretical chemist H.C. Longuet-Higgins, as well as some of the brightest talents of the rising generation of computer scientists. With Longuet-Higgins, Michie established an influential program of industrial robotics.

In 1967, Michie was given a personal professorship in machine intelligence and the unit was given departmental status as the Department of Machine Intelligence and Perception, with chairs for the senior staff. This turned out to be a mixed blessing, as it put his department into conflict with the recently formed Department of Computer Science, with which it had, at times, antagonistic relations. Michie also found it difficult to work with star academics who favoured different approaches: Gregory left for Bristol University in 1970 and Longuet-Higgins broke away to form his own research unit.

The disharmony in the Edinburgh research community, which was mirrored elsewhere, provoked the funding authorities to commission a report from Professor Sir James Lighthill FRS. The 1973 Lighthill Report was critical of the lack of progress in the field and the outcome was a massive drop in funding that cast a pall over artificial intelligence research for a decade. Within Edinburgh University, there was concern at the academic squabbling and a restructuring was imposed. A new Department of Artificial Intelligence was established with Bernard Meltzer (a former engineer) as its head and Michie was given his own independent Machine Intelligence Research Unit; but he had been frankly sidelined.

Thus, from the mid 1970s, Michie's star was on the wane. He continued his personal researches and still organised the Machine Intelligence Workshops. He accepted several prestigious visiting professorships, took up scientific journalism and continued to be the most prominent advocate for his field. In the 1980s the field recovered from the gloom of the Lighthill Report and Michie became active in the new field of expert systems.

As Professor Emeritus from 1984, he remained highly active, if less influential. He was chair of the Turing Trust which he had set up in 1975 to establish an archive for Turing's papers and a series of memorial lectures. He founded the Turing Institute in Glasgow in 1984 which conducted industrially oriented machine intelligence research for several years. In 1994 he established the Human-Computer Learning Foundation, a charitable educational trust. He was the recipient of numerous honorary degrees and achievement awards of learned societies in computing and artificial intelligence.

Michie and his former wife Anne McLaren were killed in a motoring accident while travelling from Cambridge to London last Saturday.

Martin Campbell-Kelly