Exploring Magic in the Mind with AI

  • Prof. Justin Brody is currently working on a project to enable artificial intelligence agents to recognize their own bodies and actions in a grounded way. Prof. Justin Brody is currently working on a project to enable artificial intelligence agents to recognize their own bodies and actions in a grounded way.
  • Prof. Justin Brody is currently working on a project to enable artificial intelligence agents to recognize their own bodies and actions in a grounded way. Prof. Justin Brody is currently working on a project to enable artificial intelligence agents to recognize their own bodies and actions in a grounded way.
  • A deep learning architecture for recognizing an agent's body and actions.  The network H extracts features from the agent's visual field.   The network P tries to make a prediction of the first high-level feature output by H.  The idea is to train both H and P (in tandem) so that a chosen high-level feature at the next timestep can be predicted based on the current visual field and action. A deep learning architecture for recognizing an agent's body and actions. The network H extracts features from the agent's visual field. The network P tries to make a prediction of the first high-level feature output by H. The idea is to train both H and P (in tandem) so that a chosen high-level feature at the next timestep can be predicted based on the current visual field and action.

“The mind has fascinated me for as long as I can remember.   It has long seemed to me that if there is magic anywhere in this world, there is magic in the mind,” says Assistant Professor Justin Brody.  

“As an undergraduate I began to wonder about the possibility of understanding the mental through mathematics.  I discovered the discipline of artificial intelligence and began to read as much about it as I could,” the computer scientist explains.  “Around my second year of college I read Hofstadter’s Godel, Escher, Bach and found much that inspired my later academic pursuits.  I first learned about Zen Buddhism from this book which led me to a Buddhist monastery after college; I of course also learned about Godel’s Theorems and ended up pursuing mathematical logic in graduate school.    This gave me a foundation to explore the mind phenomenologically and a deeper understanding of the mathematics that undergirds much of AI.” 

 

“The mind has fascinated me for as long as I can remember.   It has long seemed to me that if there is magic anywhere in this world, there is magic in the mind.”

 

More recently, Prof. Brody studied with Prof. Don Perlis at the University of Maryland, whose interests lie in artificial intelligence, robotics, formal commonsense reasoning and cognitive modeling.  Prof. Brody spent a lot of time discussing an AI’s model of itself, and this interaction brought together a number of the his passions.  “The idea of self is central in Buddhist philosophy, and there’s an intuition there that the self is the one idea which creates all others.   This is one of my fundamental lines of inquiry – if an artificial intelligence has a robust model of itself, how will that inform any other concepts it creates?”  The professor’s research so far has focused on bodily models of self – getting an AI to recognize its body and actions.   His long-term vision is to bring higher-level cognition into this modeling so that agents with robust notions of themselves can be created.

“My dream at F&M is to have an interdisciplinary team of students with whom I’ll work.  Computer science is wonderful in that it really allows for students to contribute in a variety of ways, including exploring conceptual ideas, coding up models, and analyzing and visualizing results,” adds the computer scientist.

Prof. Brody is currently working on a project to enable agents to recognize their own bodies and actions in a grounded way.  As mentioned earlier, he would like this to be a foundation for robust self-models for agents that can reason about their own thoughts and actions in a way that is based on their experiences.  He’s working on a number of technical projects that relate to this:  having agents learn what thoughts to focus on from their environment; exploring what concepts and representations mean for artificial intelligence; and a form of logic with a special kind of temporal self-reference.

The computer scientist has worked with collaborators at the University of Maryland Institute for Advanced Computer Studies, Bowie State and Temple University. At F&M, Prof. Brody anticipates ultimately work with students and faculty in Philosophy, Psychology and Religious Studies.

Prof. Brody received his master's degree and Ph.D. at the University of Maryland .  He previously served as an Associate Professor of Mathematics and Computer Science at Goucher College, and as a Visiting Assistant Professor in Mathematics for F&M.

His personal interests and activities outside F&M include meditation and interest in Buddhist philosophy.