Event Details

REMINDER:
If you do not have access to the 11th Floor, please check in at the main reception desk on 10th floor and someone will escort you to the conference room location prior to the start of the talk.

Meeting hosts only admit guests that they know to the Zoom meeting. Hence, you’re highly encouraged to use your USC account to sign into Zoom.

If you’re an outside visitor, please provide your: Full Name, Title and Name of Workplace to (aiseminars-poc(at)isi.edu) beforehand so we’ll be aware of your attendance.

https://usc.zoom.us/j/7042850182?omn=94494971052
Meeting ID: 704 285 0182

In 2019, DARPA funded the Machine Common Sense (MCS) program to improve AI’s “common sense,” hoping to advance the goal of producing AGI. One strategy implemented in the program involved pursuing Turing’s (1950) suggestion that it could be fruitful to attempt a simulation of a child’s mind rather than an adult’s mind. In the MCS program, this strategy entailed taking insights from empirical studies of infant development and applying them to AI research. By assembling interdisciplinary collaborative teams of developmental psychologists and computer scientists, the program encouraged the design of AI systems with conceptual competences similar to those that characterize infants and toddlers. One of the teams was tasked with evaluating these systems’ performances. Leveraging tools devised by psychologists who have studied infant cognitive development since the 1960s, the evaluation team built a battery of controlled experiments to evaluate AI systems using tests of common sense about objects, places, and agents. In this talk, I will focus on the nature of the tests that were created as well as the distinctive value that the experimental tools of behavioral science can bring to AI evaluation. In addition, I will discuss weaknesses of current AI that were highlighted by the MCS program and the unique challenges facing those working to produce AGI.

Speaker Bio

David S. Moore is a professor of psychology at Pitzer College and Claremont Graduate University. He received his Ph.D. in developmental and biological psychology from Harvard University. A developmental cognitive neuroscientist with expertise in infant cognition, Moore’s empirical research has produced numerous publications on the development of mental rotation in infancy, on the electrophysiological measurement of covert attention in infants, on the emergence of infants’ nascent “mathematical” competence, and on the value of developmental science for improving artificial intelligence. His theoretical writings have explored the contributions of genetic, environmental, and epigenetic factors to human development. Moore’s book The Dependent Gene has been widely adopted for use in undergraduate education and was nominated for the Cognitive Development Society’s Best Authored Volume award. His book The Developing Genome won both the William James Book Award and the Eleanor Maccoby Book Award from the American Psychological Association (APA), recognizing a book expected to have a profound effect on developmental psychology. Dr. Moore has served on the consulting editorial board for Child Development Perspectives and has been the editor of special issues of New Ideas in Psychology, Infancy, Developmental Psychobiology, and Infant Behavior and Development. From 2016 - 2018, he served as the Director of the U.S. National Science Foundation’s Developmental Sciences Program, and he was elected a Fellow of the American Psychological Association in 2021. He is currently a Fellow at Stanford University’s Center for Advanced Study in the Behavioral Sciences.