AIR Studies How Machines Learn and Think

Researchers from Boston University’s Artificial Intelligence Research Initiative (AIR), housed at the Hariri Institute, attempt to understand how computers make decisions once they have been programmed to learn.

Kate Saenko, an Assistant Professor in the Computer Science department at BU, takes a look into how computers programmed with artificial intelligence gather information and go about making decisions. Using both humans and computers, with funding support from the Defense Advanced Research Projects Agency (DARPA), the AIR team studies the neural pathways that evolve in artificially intelligent computers and what leads them to make certain choices. As our world becomes more reliant on AI, it becomes important to be able to trust our computers and hold their choices accountable. Kate highlights the challenges associated with understanding a computer’s learning pathways – free of human biases – as well as the need to be able to explain those decision-making pathways in a way that humans can understand.