Looking back to the future of Edinburgh’s AI legacy

By Rhona Crawford, PR and Media Manager, Communications and Marketing

In 1963, Edinburgh scientist Professor Donald Michie forecast that “in the future, people will draw information from computers as they do water from a tap.”

A quick glance at how we manage our health, energy, finances, time and even our music and TV choices gives Professor Michie’s prediction a prophetic hue. Our daily lives are flooded with data.

And the taps are being ever more loosened. Reports of new apps, chat-bots or virtual assistants are commonplace, such as the vaunted ChatGPT, which could potentially write this article one day.

Remarkable discoveries

Underpinning all of this life-changing technology is the computer science and artificial intelligence (AI) that enables our widgets to perform their magic.

Professor Michie had a ringside seat to see his prediction come true. Along with teammates Richard Gregory and Christopher Longuet-Higgins, who joined his team in the late 1960s, Edinburgh developed some of the first and most important discoveries in AI in Europe.

60^th anniversary

As the University celebrates 60 years of computer science and AI – with a year-long programme of events – we take a closer look at some of the key breakthroughs, the discoveries now embedded into everyday life and the people behind them.

The concept of AI – the science of training computers to engage in human functions such as problem solving, decision making and learning – trundled from the realms of science fiction, as seen in the 1927  film Metropolis, into nascent reality after the Second World War.

In 1963, when Harold Wilson called for a new Britain forged in the ‘white heat’ of technology, Professor Michie was assembling Europe’s first AI research hub in the University’s Hope Park Square.

 

Before long, Michie – a member of one of the code-breaking groups at Bletchley Park during WWII – was making some of the first and most important discoveries in machine learning (Michie’s early term for AI), robotics and computer games.

His obsession with the idea that machines could learn was sparked during his time at Bletchley Park, following many discussions and chess games with his colleague and friend, the late Alan Turing.

MENACE

In 1961, Donald Michie built MENACE (Machine Educable Noughts And Crosses Engine), a machine capable of learning to be a better player of Noughts and Crosses than its opponent.

As computers were less widely available at the time, Michie built MENACE from 304 matchboxes and glass beads, with each matchbox representing a single possible layout of the grid in play.

What Michie demonstrated with MENACE was the ability to build a system that could learn. Using the glass beads, he rewarded the system for good outcomes and penalised it for bad outcomes. Over the course of hundreds of games, MENACE eventually learned to play good outcomes.

MENACE's maiden tournament against Prof Michie successfully demonstrated an early form of AI in its games strategy.

This won him a bet with a colleague who believed such machines were impossible, as well as an invitation from the US Office of Naval Research to work at Stanford, where he wrote a MENACE-based program for an IBM computer.

This heralded the start of using reinforcement learning techniques to solve computing problems relating to robotics and manufacturing.

Freddy I and II

By 1973, Donald Michie and his team – which at this point included other robotic pioneers such as Robin Popplestone, Rod Burstall and Anna ‘Pat’ Fothergill - had developed a highly-teachable machine which Michie believed, “surpassed even the most versatile offerings of the robotic world of today”.

It was named FREDERICK, which stood for 'Friendly Robot for Education, Discussion and Entertainment, the Retrieval of Information and the Collation of Knowledge’. Or, to its friends in the lab, it was affectionately known as Freddy.

The team developed a pair of robots, Freddy I and Freddy II, that could identify, select and assemble objects from a jumbled heap. Michie forecast that within a decade robots could be doing boring production line jobs in an entire factory. And indeed they were.

Michie established Edinburgh’s Department of Machine Intelligence and Perception in 1966 and worked there until 1985, when he left to found The Turing Institute in Glasgow, later returning to Edinburgh as Emeritus Professor of Machine Intelligence until his retirement.

 

Multi-user systems

In the same breakthrough year – 1963 – Professor Sidney Michaelson, the University’s first professor of Computer Science formed Edinburgh’s pioneering Computer Unit.

Ironically, the Unit had no computing equipment of its own initially, so Michaelson instead relied upon a phone line to the Atlas computer in Manchester.

Michaelson’s groundbreaking research led to the development of Edinburgh’s Multi-Access System (EMAS) – a multi-user operating system, which emerged in the early 1970s. It provided the bedrock for the University’s main computing services over the next two decades.

His interests within computer science were wide ranging. He took a particular interest in the University Library and was one of the most enthusiastic advocates for its automation program, launched in 1983.

His ability to attract outstanding researchers led to his department becoming one of the strongest in Britain.

Michaelson remained at Edinburgh as Professor of Computer Science until his death in 1991 and is remembered fondly as a lively character and a generous teacher.

Software development

Professor Robin Milner joined the University in 1973 and he later co-founded the pioneering Laboratory for Foundations of Computer Science.

His research led to the development of Meta Language (ML) – a general purpose functional programming language this is still used in various forms and has influenced the development of numerous other computer languages.

Unreliable software was a major issue in early computing and ML helped programmers to verify, with mathematical rigour, that their programs were correct.

He had a vision of a new science of information, broader than computer science, which inspired the formation of the School of Informatics in the mid-1990s - through merging the departments of computer science, artificial intelligence and cognitive science.

 

Concurrent systems

During the 1970s, computing evolved from machines that could only execute one program at a time to concurrent systems, which could run several programs simultaneously. 

In 1980, Milner published a mathematical method for understanding concurrent systems, called Calculus for Communicating Systems (CCS).

Both ML and CCS were cited in the ACM Turing Award for Milner, computer science’s highest honour.

Before he left Edinburgh in 1998 to take the Chair of Computer Science at the University of Cambridge, Milner donated a sum of money to fund an annual lecture in Computer Science, which continues this year as part of the 60 year celebrations.

Legacy

Six decades after computer science and AI were established at Edinburgh, the School of Informatics is now the largest research and teaching establishment of its kind in the UK and is a worldwide center of expertise. Professor Jane Hillston, the Head of School, is extremely proud of Edinburgh’s pedigree and status. 

“Computer science and AI from Edinburgh have had a significant impact on the world through our research and continues to do so,” she says.

“In the last decade digital technology has progressed to pervade all aspects of our lives and our research and training in these disciplines are of increasing importance.

“Our 60^th anniversary also encourages us to look forward and contemplate the future. Thus we have much to celebrate but also much to do!”

 

For 60 years, Edinburgh has led many of the advances in computer science and AI that have shaped our modern world. 

Professor Mirella Lapata, Personal Chair in Natural Language Processing at the School of Informatics believes this legacy and expertise will continue to shape the course of its development for years to come.  “It is very exciting to work in field where dreams can come true,” she says. “To consider that some of the recent AI technological advancements have originated from Edinburgh is truly remarkable.

“And while the future of AI is constantly in flux, Edinburgh has remained a constant presence in it. Our AI research will keep us and our students busy in the coming future.”

Edinburgh scientists are also leading a major new research programme to expand the University’s world-leading expertise in quantum computing as Professor Elham Kashefi, Personal Chair in Quantum Computing at the School of Informatics, explains. "Our newly established quantum software lab, is building upon 60 years of tradition in classical computing to pave the way for the future of quantum computing.

“By bridging the gap between theory and practical design through direct physical implementations or simulations, we are advancing the development of quantum software theory and unlocking new frontiers of research.

“Our efforts are not only shedding light on what really matters in the field, but also pointing towards innovative research questions that will shape the future of computing. Through our work, we are poised to make a significant impact on the world by unlocking the vast potential of quantum computing.”

 

_{Picture credits: AI in the home - Oscar Wong, Getty images; Professor Donald Michie - AIAI, the University of Edinburgh; MENACE - Trial & Error by Donald Michie; Freddy II - The University of Edinburgh; Early computer model - Three Lions, Getty images;  School of Informatics at dusk - Yao Fu.}