I am eager to engage with diverse stakeholders around complex and impactful issues related to AI Safety: from rigorous scientific questions, through complex technical aspects, to cogent issues of actual uptake and governance.

Professor Alessandro Abate

Professor Abate joins the Department of Engineering from the University of Oxford, where he is Professor of Verification and Control in the Department of Computer Science. There, he leads OXCAV – the Oxford Control and Verification research group, and earlier he served as Deputy Head of Department. At Oxford, he also co-directs the Ellison Institute of Technology CDT in Fundamentals of AI.  

Professor Abate’s research focuses on control theory, formal verification and AI, and is underpinned by interests spanning decision theory, logic and probability. The overall goal of his work is to provide formal guarantees to ever more complex AI agents, particularly in the context of safety-critical, sequential decision-making problems.

“These are exciting times to work on AI Safety, and Cambridge is a wonderful platform to advance this broad research initiative,” he said. “Cambridge’s Department of Engineering and its Chair in Control represent the perfect platform for advancing science and for impactful real-world engineering applications of modern agentic AI.”

What is agentic AI and why is Safe AI a significant challenge?

Imagine a complex engineered cyber-physical system, where physical components embedded with digital devices (AI agents) interact dynamically, possibly with humans in the loop, and which are accessed through data. Is it possible to design such complex engineering components, underpinned and enabled by modern AI, with rigorous guarantees on their deployment and on their interactions with other AI agents and humans?

Professor Abate said: “As we are witnessing an inter-disciplinary convergence between scientific areas underpinned by model-based engineering and by data-driven learning, access to information-rich data and to generative models has to be traded off with a demand for safety criticality in autonomy.

“I am eager to engage with diverse stakeholders around complex and impactful issues related to AI Safety: from rigorous scientific questions, through complex technical aspects, to cogent issues of actual uptake and governance. Breakthrough research holds broad and long-ranging technological, economic and societal repercussions.” 

Professor Abate joins the long-established Control Group in the Division of Information Engineering, which carries out broad research ranging from fundamental theory through to control applications in Engineering and the Sciences. Control Engineering is a significant component of the Engineering Tripos at Cambridge.

Professor Abate’s chair will provide senior leadership and establish a new research initiative, connecting core control aspects with areas of Computer Sciences (formal methods, automata theory, logic), towards applications in modern agentic AI.

Professor Colm Durkan, Head of the Department of Engineering, said: “Given the rise of AI from its fundamental underpinnings towards a wealth of cross-cutting applications and the evidence of emergence of unexpected behaviour from large language models (LLMs), I am very pleased to welcome Professor Abate to the Department to continue his work on verification and control of AI agents, which should have far-reaching consequences.”

The Professorship of Control Engineering was previously held by Rodolphe Sepulchre, Keith Glover and Sir Alistair MacFarlane.