Prof. Benjamin Kuipers, known for his AI research on common sense knowledge, retires

In pursuit of a mathematics for investigating the mind, he has brought us important insights into how AI agents can learn about the world from their own experience, and how ethics relates to AI.
Benjamin Kuipers portrait
Electrical Engineering and Computer Science professor Benjamin Kuipers

Benjamin Kuipers, professor of electrical engineering and computer science, will retire from the faculty on May 31, 2024, after 15 years at the University of Michigan. Kuipers is best known for his work on computational models of cognitive maps, robot exploration and mapping methods, the qualitative simulation algorithm QSIM, and foundational learning methods. He has also been an early thought leader on the subject of ethics for AI and robots.

During his 46-year career in academia, Kuipers has also served on the faculty at Tufts University, in the Mathematics Department; and at the University of Texas at Austin, in the Computer Science Department, where he also served as department chair for four years.

Kuipers was born in Grand Rapids, Michigan, the oldest of five children, and grew up here in Ann Arbor, where he graduated from Ann Arbor High School. Beginning in his middle school years, Kuipers and his father, who was an engineer, would take walks in the evening during which they would talk about mathematics, science, and the mind – and whether there would be a science of the mind and what kind of math would be needed to have a science of the mind. This fired his interest, and he couldn’t wait to get to college and take psychology courses. 

Then he got to college, and he took a psychology class. “It was a crashing disappointment.” recalls Kuipers, “The interesting parts weren’t rigorous, and the rigorous parts weren’t interesting.”  So he majored in math, which he knew and loved, and earned his B.A. with High Honors in Mathematics from Swarthmore College.

During those days of the Vietnam War, he realized that he was opposed to all war, and the draft board classified him as a conscientious objector to military service.  After completing his bachelor’s degree, he performed two years of alternative national service, working as a programmer in a computer lab in – of all places – the Psychology Department at Harvard University.

Kuipers ultimately earned a PhD in Mathematics from MIT.  During his graduate studies in pure math, he took a graduate introduction to artificial intelligence and discovered the field of AI. “The skies opened,” he said.  “I realized that this is what I had been looking for all of my life.”  He saw the kinds of computational methods used in AI as a starting point for developing the math of the mind, a kind of applied math that could express clear and precise hypotheses – ones that could be implemented as running computations – about the nature of the mind and cognition.

Kuipers wanted to work on the mind, but that’s a very broad subject. He started spending his time at the MIT AI Lab, where students and faculty were working on many aspects of AI, including perception, planning, language, motion, and knowledge representation.  As he surveyed this, the part that particularly appealed to Kuipers was commonsense knowledge, and within that area, the cognitive map: how an agent, human or AI, can build a map of its environment from its own experience. 

Kuipers completed a dissertation under the guidance of Marvin Minsky entitled “Representing Knowledge of Large-Scale Space,” which was one of the first computational models of cognitive maps. This was a topic he returned to repeatedly over the decades, in the context of robot exploration and mapping, and of developmental learning.

Toward the end of a DARPA-funded post-doctoral year, Kuipers discovered that the primary interest in his work on cognitive maps came from military agencies with the goal of building intelligent cruise missiles.  He knew that he did not want his life’s work to contribute to war and has not taken military funding since.

He took a faculty position at Tufts University teaching math and CS. While there, he joined a research group doing AI and medicine at MIT and collaborated with a colleague at Tufts Medical School on how expert physicians reason qualitatively about the physiological mechanisms of the body, both cognitively and computationally. 

Kuipers moved to the University of Texas at Austin and built a flourishing research group there. Over 23 years at UT he established himself as a leading AI researcher, and graduated 30 PhD students. During four of those years, from 1997–2001, he served as department chair.

Ram Ramamoorthy, the Personal Chair of Robot Learning and Autonomy in the School of Informatics at the University of Edinburgh, was one of Kuipers’ students at UT Austin. He reflects fondly on his time as Kuipers’s student, “because Ben created a welcoming and nurturing environment. Even when the scientific activity was highly competitive, the interpersonal dynamics were that of a team of people that cared about each other.”

“Another attribute that I enjoyed is the intellectual humility and curiosity to learn about new things. Ben had a strong sense of a personal research agenda, but he also allowed his students to explore to develop their own research personality. This included, on occasion, situations where the students would become expert in something Ben didn’t know much about, and he was perfectly happy then to learn and explore new areas.”

Over the years, Kuipers developed some of the earliest computational models of the cognitive map: the representation and learning of knowledge of navigational space. He developed robot exploration and mapping methods based on these models. He also developed the QSIM representation and algorithm for qualitative simulation of dynamical systems, which was mathematically elegant and which supported sound qualitative reasoning about a variety of interesting physical systems. It had a great influence on the field of commonsense knowledge representation. And, he has worked on simulated and physical robotic models of developmental learning of the foundations of commonsense knowledge. 

After joining the University of Michigan, Kuipers played a leading role in establishing the Robotics Institute, which later became the Robotics Department in the College of Engineering.  So far at the University of Michigan, he has graduated four more PhD students.

In this video, Prof. Kuipers discusses how advances in AI and robotics have raised questions about the impact on our society of intelligent robots, unconstrained by morality or ethics.

Collin Johnson, Staff Autonomy Engineer at May Mobility, was one of Kuipers’ students at Michigan. He said, “When I think of Ben, I think of wisdom. At a recent gathering of his PhD students from across the years, every single student had some key phrase that they have carried with them over the years. In my post-PhD career, I often find myself paraphrasing Ben when explaining general principles of robotics to younger engineers.”

Johnson adds, “Many people have commented to me at conferences about Ben’s cohesive vision for how to understand the world. Ben’s view of the world is formed from a life dedicated to research and reflection. His combination of curiosity and conviction inspire me to do my best to follow my own path and avoid simply doing whatever everyone else is doing. Even when it is difficult to see where that path will lead, Ben provides a shining example that making your own path can lead to a rich and full life.”

While at Michigan, Kuipers became an early and leading voice on the topic of ethics in the fields of artificial intelligence and robotics. In 2018, he authored an article in the Communications of the ACM entitled “How Can We Trust a Robot?”, which discusses the importance of trust in relationships, and why trust and ethics in AI and robotics will be essential to the successful continued functioning of society.

In 2020, Kuipers introduced the course “Ethics for AI and Robotics.” In 2022, he organized the Obert C. Tanner Lecture on Artificial Intelligence and Human Values at the University of Michigan.  He plans to continue reading, writing, and speaking on AI, ethics, and the mind for the foreseeable future.

Kuipers is a Fellow of the American Association for the Advancement of Science (AAAS), a Fellow of the American Association for Artificial Intelligence (AAAI), a Fellow of the Institute of Electrical and Electronics Engineers (IEEE), and a Fellow of the Asia-Pacific Artificial Intelligence Association (AAIA). He has been recognized by the College of Engineering with the Herbert Kopf Service Excellence Award, and by the EECS Department with an Outstanding Achievement Award.