Department of Machine Intelligence and Perception, University of Edinburgh

Beginnings in a flat

Edinburgh's artificial intelligence work started small. Donald Michie, then Reader in Surgical Science at the University of Edinburgh, formed a small research group into artificial intelligence in a flat at 4 Hope Park Square in Edinburgh. Michie's route into computing ran through the Second World War: he had been a member of Max Newman's code-breaking group at Bletchley Park, and working with Alan Turing he was introduced to computing.^[1]

From a unit to a department

The university soon recognised artificial intelligence as a field worth a formal home, and in January 1965 it established the Experimental Programming Unit, with Michie as Director. In October 1966 that unit was transformed into the Department of Machine Intelligence and Perception. Richard Gregory and Christopher Longuet-Higgins were also instrumental in setting up the new department.^[1]

The name itself is telling: it put machine intelligence alongside perception, so that building machines that could act in the world and understanding how perception works were treated as parts of one enterprise. Christopher Longuet-Higgins went on to take a leading role in starting the School of Epistemics at Edinburgh, and he coined the term cognitive science in a paper in 1973 — a sign of how closely early AI at Edinburgh was tied to the study of minds.^[1]

Early achievements

Edinburgh's early AI researchers produced results that are still remembered. One was FREDDY, a robot able to assemble objects automatically from a heap of parts; it is now held in the National Museum on Chambers Street. On the software side, Robin Popplestone and Rod Burstall designed and developed the POP-2 symbolic programming language at Edinburgh, and David Warren designed and developed the Edinburgh Prolog programming language.^[1]

Both of these are symbolic programming languages, meaning that the main objects they manipulate are symbolic expressions rather than numbers — the natural choice for programs that reason about structures, rules and relationships instead of computing numerical results. For a department whose name promised machine intelligence, tools of this kind were the working material: a robot that could pick parts out of a heap and put them together needed ways to represent the parts and the task, not just ways to calculate.

Applied work and merger

In 1983 Jim Howe established the Artificial Intelligence Applications Institute, which was meant to expand applied AI work between academia, industry and government, taking the department's research beyond the laboratory.^[1]

In 1998 the Department of Artificial Intelligence, the Centre for Cognitive Science, the Department of Computer Science, the AI Applications Institute and the Human Communication Research Centre merged into the School of Informatics.^[1] That merger gathered Edinburgh's AI, cognitive science and computing groups under a single school, and it is the form in which most of this work is organised today.

A wider computing context

Edinburgh's computer science work itself dates back to the early 1960s, when a Computer Unit was created under Sidney Michaelson, who became the university's first Chair of Computer Science in 1967.^[1] The AI group therefore grew up alongside, and eventually together with, a broader computing effort at the same university.

Seen from the outside, the department's significance lies in timing and in institutional form. Within a few years an informal group meeting in a flat became a named department with a director, a research programme and its own programming languages and robots — one of the ways artificial intelligence moved from a topic discussed by individuals into a recognised academic subject with buildings, staff and students. The later merger into the School of Informatics carried that history forward rather than ending it, and the survival of FREDDY in a public museum is a reminder of how early some of this work was.