The Promise and Peril of Artificial Intelligence
The Promise and Peril of Artificial Intelligence
Part 3 of the 2020 Arizona Science Lecture Series
Speakers: Stephen Kobourov, Carlos Scheidegger
The word robot is 100 years old, but only recently has AI begun to make real-life impact, from Apple’s Siri to Uber’s self-driving cars. Rapid advances in machine learning have renewed the idea of modeling how the human brain works by building deep neural networks that learn how to solve problems with the help of many examples. Like other revolutions, AI comes with great promise: better medical diagnoses, more efficient transportation, and personalized recommendations from shopping to music to fitness routines. There’s also peril, since AI enables mass surveillance and manipulation, and perpetuates societal biases. There are technological challenges—deep neural networks can solve only narrow problems, are not robust, and do not generalize how we expect them to—but a truly humane, AI-enabled future will require much more than just technologists. We must work with ethicists, policy-makers, and particularly the people that will be affected by these systems.
