In this talk, Nick Haber will present a series of projects aimed at realizing a symbiotic relationship between early human learning and artificial intelligence. Young children are experts at playing, an ability to explore and structure their environment and interact with others in ways that are widely thought to teach them a great deal about the world, quickly. Yet our understanding of this remarkable way that children learn remains imprecise. How might we, for instance, design artificial intelligence systems that learn in this way? While artificial intelligence has made dramatic gains in the past decade, training these systems requires large amounts of carefully curated and labeled data, a need that contrasts strongly with the exploratory processes of children. Not only would the answer to this question be a great leap forward in the design of artificial systems, but it also holds the potential to teach us more about ourselves. By building computationally-precise models of early learning, we hope to better understand how these building blocks contribute to learning throughout life. Further, if we are able to understand how early learning varies, we can design new learning tools that better serve learning differences.
To realize this, Nick uses modern tools in deep reinforcement learning, wearable devices, cognitive models, and human subject research. First, he presents on the design and clinical validation of a computer vision-powered learning tool on Google Glass for children with autism, an effort that came to be known as the Autism Glass Project. Next, he describes results on artificial intelligence modeling of early human learning and how that can both help us better understand how we learn and how to develop tools that help us learn better. Finally, Nick turns to work on how people attend to animate stimuli, including preliminary results on fundamental learning differences for people with autism.
Nick is a postdoctoral fellow at Stanford University. His work bridges artificial intelligence, cognitive modeling, and wearable device development. After receiving his PhD in mathematics, he launched the Autism Glass Project, developing and testing a computer vision-powered learning tool on Google Glass for children with autism. His recent work includes the design of artificial intelligence systems aimed at increasing our understanding of early childhood learning.