Edinburgh and AI: sixty years of teaching machines

AI research at Edinburgh began in 1963, when Donald Michie set up a small research group at 4 Hope Park Square. The University now calls the city “the birthplace of AI in Europe”. In the sixty-odd years since, that group grew into a department and then a school, built one of the first robots to assemble objects using vision, and came through Britain’s first “AI winter”. Here’s the short version, and why it still matters if you run a small organisation here today.

Before computers

Long before anyone said “artificial intelligence”, Edinburgh produced people who thought hard about numbers and the physical world. John Napier, born at Merchiston Castle in 1550, published his logarithms in 1614. In 1617 he described “Napier’s bones”, numbered rods that let you multiply by placing them side by side and reading off the answer. James Clerk Maxwell was born at 14 India Street on 13 June 1831, and went on to do landmark work on electricity, magnetism, optics and the kinetic theory of gases.

Neither worked on anything like AI, so we won’t claim them. They’re simply a reminder that careful mathematical thinking has deep roots here.

1963: a small group at Hope Park Square

Michie had worked on wartime code-breaking at Bletchley Park, and was a Reader in Surgical Science when he set up his group. It became the Experimental Programming Unit in January 1965, with Michie as director. In October 1966 it grew into the Department of Machine Intelligence and Perception, funded by a large Science Research Council grant held by Michie, Christopher Longuet-Higgins and Richard Gregory.

The University says that “since 1963 the University of Edinburgh has been at the forefront of AI”, having established “Europe’s first research group” and been “the second university in the world after Stanford to teach the subject”.

Michie also had a good eye for where things were heading. In 1968 he predicted: “Along with question-answering services, which will allow us to inquire about the restaurants in our locality or politics in Paraguay, will come the games opponent, the puzzle setter and the quiz master.” Anyone who has asked a chatbot for a lunch recommendation will recognise the first part. (For how today’s versions work, see our plain-English guide to large language models.)

FREDDY, the robot that took 16 hours

Michie’s department built two robots: FREDDY (1969 to 1971) and FREDDY II (1973 to 1976). FREDDY II was a stationary robot with a moving television eye and a pincer hand. Given a jumbled heap of parts for a wooden toy car and boat, it could use vision to pick out the right pieces and assemble both, in about 16 hours.

The team had to design and build the robot, its programming system and its vision system from scratch. The slowness wasn’t the point: the robot moved slowly because of the limited computing power of the day.

Winter, then a department

Not everyone was impressed. In the early 1970s the mathematician James Lighthill reviewed AI research for the Science Research Council, and he was especially critical of robotics. The council sided with him and cut funding for AI and robotics, a decision IEEE Spectrum says “ushered in the first British AI winter”. By its account, Michie’s lab was left with Michie, one technician and an administrative assistant.

Edinburgh kept going anyway. The Department of Artificial Intelligence was formed in 1974. The Artificial Intelligence Applications Institute (AIAI) followed in 1984, and its name tells you what it was for. In 1998 the Department of AI, the Centre for Cognitive Science and the Department of Computer Science came together in what’s now the School of Informatics.

Edinburgh today

Since 2014, new institutions have joined the old ones:

• The Data Lab, “Scotland’s Innovation Centre for Data and AI”, was founded in 2014. It’s hosted by the University of Edinburgh at the Bayes Centre, with hubs in Glasgow and Aberdeen.
• The Bayes Centre opened in October 2018 as part of the Edinburgh and South East Scotland City Region Deal. The University of Edinburgh and Heriot-Watt’s £661m Data-Driven Innovation initiative, also part of the deal, aims to “support Edinburgh in its ambition to become the data capital of Europe”.
• The Edinburgh Futures Institute opened in the old Royal Infirmary in June 2024.
• The UK’s national supercomputer will be hosted by EPCC at the University of Edinburgh, backed by up to £750m from the UK Government, confirmed in June 2025. Construction began on 25 June 2026.

Why this matters to a small business

Being near a famous university won’t make your AI project work. But three things are genuinely useful.

Talent. Edinburgh’s universities train people in AI and data every year, and some of that talent is easy to reach. The Data Lab’s MSc Placement Programme puts students on 8 to 12 week projects, with the host paying the Real Living Wage plus on-costs.

Support. Free advice is close by, from The Data Lab, Business Gateway and the national AI adoption programme they help deliver. Our guide to paying for an AI project in Scotland covers what’s open.

A habit of careful thinking. FREDDY’s 16 hours and the winter that followed Lighthill’s report carry a lesson for anyone buying or selling AI: real progress is slower than the hype, and overreacting in either direction is expensive. The useful question isn’t “is AI amazing?” but “does this tool do this job reliably, and how will we check?” The old Scots advice still holds: “ca’ canny”, go carefully. It’s how we work, and you can read more about us.