Dr. Crispin Barker: Salami Tactics and the AI Control Problem: The Case of Autonomous Weapon Systems

​Dr. Crispin Barker will report on his research into the history of salami tactics (a strategy associated with the Cold War but with much older antecedents) and how an advanced artificial intelligence could use salami tactics to subvert control. His research explores the fundamental flaws in human decision-making that have made even highly experienced politicians and military officers vulnerable to salami tactics, and shows how these can be exploited by an autonomous weapon system to negate the means of control without being shut down or destroyed. His report concludes by discussing how humans can use salami tactics against AI to maintain control.

​Dr. Crispin Barker is a historian and philosopher of science, medicine, and related domains. He received his Ph.D. in the History of Science and Medicine from Yale and has lectured at the University of California, Berkeley, the University of Zurich, and the University of Lucerne. He is currently a Research Associate at the Center for Science, Technology, Medicine & Society at UC Berkeley. In recent years he has worked on the philosophy of artificial intelligence with a particular focus on autonomous weapon systems, how to mitigate the risks of advanced AIs, and the feasibility of using AI instead of people to solve problems.