LTT: B. Karlan

Human Achievement and Artificial Intelligence

Brett Karlan, University of Pittsburgh, HPS

ABSTRACT: Shortly after the deep neural network AlphaGo defeated Go grandmaster Lee Sedol in a series of matches in 2016, Sedol announced his retirement from professional Go. In his retirement announcement, Sedol expressed a sentiment many others have echoed about the encroachment of AI technologies into spaces of important human value, arguing that the success of AlphaGo made it pointless for human beings to continue trying to be perform at high levels in Go. In this paper, I ask: is this worry justified? What is valuable about difficult human achievement, and why might such value be threatened by superhuman performance levels of AI technologies in those areas? I argue that, on several plausible accounts of the value of achievement, Sedol’s worries are misplaced. Much like a sprinter’s achievement are not threatened by the performance of a supersonic jet, human cognitive achievements are similarly insulated from losing most of their value in an AI-dominated future. Nevertheless, the threat seems to be psychologically salient to many, and this fact means some aspects of the theory of human achievement might have to develop a self-effacing character. I show how this can be done in this paper, as well as try to catalogue more thoroughly the underlying sources of this (what I think of as an ultimately misplaced) worry.

You have the option to attend in person, watch a YouTube livestream, or watch through Zoom.

The Zoom registration link is:  https://pitt.zoom.us/webinar/register/WN_rEwkdh9qTTyhTzl5Gh5BEQ

Please Note: Non-Pitt individuals who want to attend our in-person talks must send an email in advance to Katie Labuda (kathleenlabuda@pitt.edu) requesting Guest Building Access, or you will not be able to enter the Cathedral of Learning.