Prof. John Laird retires after a 36-year career in artificial intelligence at Michigan

He is the architect of Soar, one of AI’s most enduring general cognitive architectures.
John Laird photo
John Laird, outside the Bob and Betty Beyster Building.

John E. Laird, the John L. Tishman Professor of Engineering, retired from the EECS faculty on May 31, 2022, after 36 years at the University of Michigan. His career has been marked by his groundbreaking research into cognitive architectures, his work in AI course development and teaching,  and his service to the department.

John was born in Ann Arbor, where he attended Huron High School. It was there, in 11th grade, that he was first exposed to computing. His math teacher had a relationship with a local company that had computers, and as a result Huron had a teletype station installed in the math area, and students were able to write programs.

John enjoyed computing, and upon beginning his college studies at U-M started taking computer classes for fun while concentrating on math, which he had been good at in high school. “Eventually,” he said, “I had an epiphany and realized that I liked the CS classes a lot more than the math, so I decided to go with what I really enjoyed.”

While at U-M, John earned the titles of Regent Scholar, Branstron Scholar, and Angell Scholar, and eventually a degree in Computer and Communications Sciences in the College of LSA, in 1975. 

After receiving his bachelor’s, John went to work for Burroughs, where he worked on the compiler for the S1000 check sorting system. Although he enjoyed working at Burroughs, “it convinced me to go to graduate school,” he said.

John attended Carnegie Mellon University for graduate school, where he had the great fortune to meet with two people with whom he would form lasting and impactful relationships. Allen Newell, John’s advisor, was one of the giants of AI. His guidance was critical to John’s success as a graduate student and beyond. In addition, John met another student, Paul Rosenbloom, who shared John’s goal of exploring the architecture underlying the mind. 

John and Paul both joined Newell’s Instructable Production System project, the goal of which was to build very large rule-based systems through instruction. That project failed, but from it, both John and Paul learned a lot about building AI architectures.

John picked up on a collection of ideas from Allen Newell on alternative approaches to organizing rule-based systems in terms of problem spaces, which in turn led to John’s research direction in cognitive architectures. “Cognitive architecture was not a thing when I started,” said John, “but Allen Newel, Paul Rosenburg and I had this common interest, and so that was the direction we started going.”

John called his architecture Soar, and its first release was in 1983, the year he completed his PhD. On March 25, 1986, John came back to Ann Arbor as a faculty candidate. His seminar was entitled, “Soar: An Architecture of General Intelligence.” 

Since joining the U-M faculty in 1986, John has continued to do research on the architecture of the mind, developing and evolving the Soar architecture and applying it to new and ever more challenging domains. 

In 1998, he helped found Soar Technology, Inc. along with some of his research staff from Michigan. Soar ​enables the creation of autonomous software agents that can assess large amounts of human knowledge and put that knowledge to work assisting human decisions. John is currently chair of the board for Soar Technology.

In 2012, John published a book entitled “The Soar Cognitive Architecture,” and in 2018, John and his longtime collaborator Paul Rosenbloom received the Herbert A. Simon Prize for Advances in Cognitive Systems for the pair’s research on cognitive architectures, especially the Soar project.

Most recently, working with his students, John has extended Soar to include reinforcement learning, episodic memory, semantic memory, mental imagery, and emotion-inspired processing, leading to AI systems that learn new tasks interactively from natural language instruction.

Today, the Soar architecture is in use all over the world, for research on artificial intelligence and cognitive science, and finding a variety of applications across no end of domains. John held the 42nd Soar workshop on May 24–25, 2022.

In the late 1990s, John began exploring computer games and AI. This involved investigating opponent modeling and the use of a computer game to develop an advanced AI. He and his students developed Haunt 2, a mod to Unreal Tournament where human-level AI characters really make a difference. They also developed a Soar Quakebot that was able to anticipate opponents’ actions, using the same technology as that the researchers had used for simulating military pilots. 

In parallel with this effort came a course, EECS 494, Computer Game Design and Development, which culminated each term in a very popular Computer Games Showcase. John taught this course through 2012, when a faculty dedicated to video game design was hired. The course, now called Introduction to Video Game Development, remains popular to this day.

While teaching the course, John tapped his former student Sid Meier, the developer of many groundbreaking video games including the Civilization series, to give an annual lecture for students interested in the game industry. 

John has advised 34 PhD recipients and is currently advising six students.

John has also made great contributions that have helped CSE to be a successful organization. From 1994–1999, he was the director of the Artificial Intelligence Laboratory. In 2000, he stepped up from lab director to become the associate chair of EECS, for the CSE Division, through 2004. John later served as interim chair of CSE in 2011. 

John has long been connected with the physical space that houses CSE. He was involved in the 2002–2006 design process for CSE’s current home, the Beyster Building (originally called the CSE Building), and the Laird family donated the funds to create the stained glass that makes its signature conference room, on the third floor of the building, so unique. Since 2018, John has been instrumental in moving the upcoming Leinweber Computer Science and Information Building from concept to reality. The Leinweber Building will provide expansion space for CSE and will co-locate the School of Information.

In retirement, John is looking forward to continuing his work in the area of cognitive architectures. He and three of his former students – Peter Lindes, Bob Wray, and James Kirk  – have formed the Center for Integrated Cognition to continue his research on AI systems that use many different cognitive capabilities for complex problems. The center already has funding from government agencies.

Regarding other plans, John and his wife plan to be more flexible and to travel a bit more. “My wife suggested that I should take up something like woodworking,” said John. “That’s not going to happen. No completely new hobbies!”

John Laird is a fellow of the Association for the Advancement of Artificial Intelligence (AAAI), the Association for Computing Machinery (ACM), the Cognitive Science Society, and the American Association for the Advancement of Science (AAAS). In 2020, he received the Stephen S. Attwood Award, the most prestigious award from the College of Engineering, for his teaching, research, and service activities.

Explore:
Division News; John Laird