The speaker discusses their reservations about using algorithms to explore definitions of fairness, and how the successes in machine learning have been limited to a narrow set of tasks.