Michael Kearns
Showing 1–4 of 4 transcripts.
- Lex Fridman8 min
Michael Kearns: Algorithmic Trading and the Role of AI in Investment at Different Time Scales
Financial markets currently utilize algorithms for high-speed execution and statistical arbitrage, yet long-term investment strategies remain resistant to full automation due to the complex, human-centric nature of navigating geopolitical shifts and economic cycles. A workshop co-sponsored by the speaker and the Federal Reserve Bank of Philadelphia highlighted that machine learning is still in its early stages for macroeconomic prediction, failing to replicate the synthesis of diverse data over decades. Consequently, experts assert that roles demanding long-term risk appetite and nuanced political judgment are secure from imminent algorithmic displacement, with no "robo Warren Buffett" emerging in the near future.
- Lex Fridman8 min
Michael Kearns: Differential Privacy
Differential privacy establishes a rigorous standard by ensuring analysis outcomes remain statistically indistinguishable whether an individual's data is included or excluded from a dataset. This is achieved by transforming deterministic algorithms into probabilistic models that inject calibrated noise, thereby preventing the reverse engineering of specific records while enabling tasks like neural network training and hypothesis testing. The framework has evolved from skepticism to practical viability, allowing the scientific community to extract valuable predictive insights from aggregate data without compromising mathematical privacy guarantees.
- Lex Fridman7 min
Michael Kearns: Game Theory and Machine Learning
This discussion explores the convergence of game theory and machine learning, where algorithms like those in navigation and social media platforms compute selfish best responses to guide users toward a Nash equilibrium. While John Nash's foundational work provides the mathematical stability for analyzing such interactions, the presentation highlights a critical limitation: achieving this competitive equilibrium does not guarantee an optimal collective outcome and can sometimes degrade overall system efficiency. The conversation concludes by examining how modern algorithmic design aims to mitigate these risks by actively steering participants away from inefficient equilibria.
- Lex Fridman1h 49m
Michael Kearns: Algorithmic Fairness, Privacy & Ethics | Lex Fridman Podcast #50
University of Pennsylvania professor Michael Kearns explores the ethical boundaries of algorithmic systems in his book *An Ethical Algorithm*, highlighting the mathematical impossibility of simultaneously satisfying all fairness metrics while advocating for Pareto curves to let policymakers visualize accuracy versus bias trade-offs. Kearns distinguishes between the rigorous mathematical guarantees of differential privacy and the flawed nature of traditional anonymization, arguing that privacy must be preserved through calibrated noise rather than data masking. Furthermore, he applies algorithmic game theory to explain how optimization for engagement on social platforms inadvertently drives societal polarization, urging a shift where human values are explicitly injected into objective functions rather than left to autonomous systems.