AI - Pick Two: Fast, Cheap, or Good?

Abstract

Assuming artificial intelligence is fast, cheap, and good is a mistake that many people make -- often hoping for the best of all worlds. Overestimating the capabilities of AI can lead to unintended or harmful outcomes. Short of a dream AI, there are good options and uses for AI technology. We argue that everyone should really think through the implications of these options, their use of AI, and how AI is being used to shape their lives. Literacy in AI is becoming a need for citizens across our society, as AI is having a pervasive impact with potential for both good and for bad. As the nation is poised to invest billions of dollars that is critically needed to remain the leader in artificial intelligence, we have to create an innovation ecosystem that is not just about technological achievement but also has an intentionality to benefit and uplift everyone in our society. The aim of this talk is to calibrate expectations such that we can make wise and effective choices in the use and development of AI. This talk will provide: 1) an overview of the space of current options in AI and its amazing rate of technological advancement, 2) systemic inequities in AI systems and the biases that can result in the use of AI, and 3) hypotheses about avenues that could lead to future dream AI systems, which have the interpretability and accountability needed to have a positive impact.