Gali Noti is a native of Yuvalim, a village in the lower Galilee. Her undergraduate work in computer science and cognitive sciences at The Hebrew University of Jerusalem focused on human decision-making processes, inspired by the classic studies in the field, led by Tversky and Kahneman, who demonstrated how mathematical, rational models fail to explain them. Realizing that these ideas had not yet entered the discourse in computer science, and that analyses in core real-world applications were still based on unrealistic behavioral assumptions, Gali decided that the subject called for further investigation.
She continued to an MSc in computer science at the Hebrew University’s Center for the Study of Rationality, where her cross-disciplinary approach was encouraged. Under the supervision of Prof. Noam Nisan (CS) and Prof. Ilan Yaniv (Department of Psychology), she demonstrated the importance of addressing behavioral features in algorithmic environments, and specifically that in the context of Internet ad-auctions, the gap between theoretical and actual behavior might be worth billions of dollars!
In her PhD dissertation, entitled “Behavioral Algorithmic Game Theory,” Gali, in a series of papers on behavior-based econometrics and behavior-based machine learning, studied strategic computational systems in which human players interact, with the goal of bridging the gap between the theory and actual human play. The results show how analyzing strategic online systems according to the standard economic assumption that humans are rational can result in significant errors that may critically impact social welfare and platform performance. The
challenge, in her view, is to appropriately incorporate in the models underlying these systems insights from behavioral theories and empirical data, and to apply these behaviorally appropriate models to improve the design of systems for human users. During her PhD studies, Gali was also involved in an applied research project at Microsoft Research.
Today, Gali’s scientific work is in a new field of research called Human-Centered Artificial Intelligence. Much work has shown that AI is able to perform complex tasks and often outperforms human capabilities. However, since in the real world these systems do not function alone, but together with real people, it is important to develop them in a way that takes into account human behavior, preferences, and needs. Gali’s goal in her postdoctoral training in the United States, is to study and develop new AI frameworks for hybrid human–AI teams.
She develops new forms of human–AI interactions that foster human–AI collaboration in decision-making, leveraging the complementary strengths of the human and of AI for the good of the human users and society. She takes an interdisciplinary approach that involves machine-learning techniques, game theory, econometrics, and experimental methods, together with insights about human decision-making behavior, to study and design human–A I systems with improved joint performance. Her results stress the importance of behavior-aware and interactive approaches to improving human–AI systems for their human users.