Science on Screen: The A.I. Alignment Problem from "2001: A Space Odyssey" to Today

Our world is abuzz with concerns about the rise of artificial intelligence and its potential impacts on our jobs, social relationships, the environment, and society. As artificial intelligence systems become more powerful, one of the core challenges is ensuring that they remain aligned with the moral values and desired outcomes of their human creators, which are often subjective, nuanced, and difficult to articulate.

This challenge is known amongst researchers as the A.I. alignment problem. Perhaps no figure better illustrates this challenge than HAL 9000, the fictional antagonist of Stanley Kubrick’s science fiction masterpiece 2001: A Space Odyssey (1968), whose unyielding execution of his pre-programmed mission has catastrophic consequences for the astronauts aboard the spacecraft.

2001: A Space Odyssey is a sprawling drama of man versus machine that begins in our prehistoric ape-ancestry past, then leaps forward millennia into colonized space, and ultimately whisks astronaut Bowman into uncharted space, perhaps even into immortality. Widely considered among the greatest films of the 20th century, Kubrick’s film was also remarkably prescient in its depiction of artificial intelligence technologies.

In this Science on Screen event, Dr. Joe Halpern and Dr. Shaun Nichols will discuss how HAL 9000 helps us understand the A.I. alignment problem. Their conversation will touch on the historical development of artificial intelligence systems since the 1950s and new ethical challenges presented by autonomous decision-making in algorithmic data models.

Free admission! This screening will be presented as part of National Evening of Science on Screen®.

Science on Screen® is an initiative of the Coolidge Corner Theatre, with major support from the Alfred P. Sloan Foundation.

About the Speakers

Joseph Halpern is the Joseph C. Ford Chair of Engineering and Professor of Computer Science at Cornell. His research focuses on reasoning about knowledge and uncertainty, and its applications to distributed computing, AI, security, and game theory. He is also interested in such topics as fault tolerance in distributed computing, modal logic, causality, program verification, and programming language semantics. His work lies at the boundary of a number of fields: for a talk at Princeton, he described himself as someone with a PhD in mathematics, who calls himself a computer scientist, giving a talk to economists about a subject mainly studied by philosophers.

Shaun Nichols is the Distinguished Professor of Arts & Sciences in Philosophy at Cornell. He works at the intersection of philosophy and cognitive science. Most of his research aims at understanding the psychological and cultural underpinnings of philosophical thought. 

Part of our "Science on Screen" series. Courtesy of Warner Bros. and Swank Motion Pictures.