Gary N. Smith and Robert J. Marks discuss the inability of AI to understand puns, lyrics, context, or anything at all. From trading futures, predicting political outcomes, and parsing lyrics, the fundamental incomprehension of artificial intelligence is a key to understanding its limitations.

Show Notes

• 02:30 | The AI Delusion by Oxford University Press
• 03:00 | The importance of knowing how to interpret data
• 03:40 | The stock market and the “Super Bowl Indicator”
• 04:00 | “Utter nonsense” and mistaking correlation for causation
• 04:30 | Computers and generating incidental correlations
• 05:00 | The quants (a quantitative analyst or data modeler) versus the analysts in investing
• 05:50 | Quants, trading futures, and self-fulfilling prophecies
• 07:00 | Obama, Clinton, “Ada”, and the effectiveness of data analytics in politics
• 10:15 | Some say, “If you can’t measure it, it doesn’t count.” Actually, often the things that count can’t be measured.
• 10:40 | Neural networks and the Bill Clinton swing vote
• 11:20 | Gaming the Watson showdown on Jeopardy
• 13:15 | The Winograd Schema Challenge
• 15:00 | IBM Watson’s complete failure to understand Bob Dylan
• 17:00 | AI’s inability to understand words or meaning

Additional Resources

• Bio and faculty page of Gary N. Smith at Pomona College
• Standard Deviations: Flawed Assumptions, Tortured Data, and Other Ways to Lie with Statistics
• The AI Delusion
• About the Winograd Schema Challenge, by Davis, Morgenstern, and Ortiz
• Bob Dylan and IBM Watson on the meaning of his lyrics