How to Study Machine Behavior
Researchers propose a multidisciplinary approach to AI
Machines powered by artificial intelligence (AI) increasingly mediate our social, cultural, economic, and political interactions. Understanding the behavior of AI systems, therefore, is essential to controling their actions, reaping their benefits, and minimizing their harms.
That’s the idea behind Machine Behavior, a proposed new area of research that would study AI beyond computer science and engineering and see its impact on biology, economics, psychology, and other behavioral and social sciences. Do machines take on their own behaviors as they gain sophistication?
In a new research paper — published in Nature on April 24, and described in a new video — a distinguished group of authors from many fields argue that an interdisciplinary understanding of AI “necessitates a broad scientific research agenda to study the ‘machine behavior’ that incorporates and expands upon the discipline of computer science.” It must include insights from across the sciences, they said. “We first outline a set of questions that are fundamental to this emerging field and then explore the technical, legal, and institutional constraints on the study of machine behavior.”
MIT IDE lead Alex “Sandy” Pentland is one of the co-authors of the paper along with his MIT CSAIL colleagues, Iyad Rahwan, Manuel Cebrian, Nick Obradovich, and Cynthia Breazeal. MIT also collaborated with a diverse set of researchers from the Max Planck Institutes, Stanford University, the University of California San Diego, and other educational institutions, as well as from Google, Facebook, and Microsoft. The paper makes the case for this new area of research aimed at understanding the behavior of AI systems and their potential.
Read more here and here and watch the video about the project here.