About the lecture
What do ancient philosophical texts from Plato and Aristotle have to do with our contemporary efforts to come to grips with AI, LLMs, and so on? Everything, I will argue.
I begin with Plato’s Phaedrus and the story of Theuth, the inventor of writing, as a “growing up” story that teaches young Phaedrus how to avoid the central risks of the then-new media technology of writing. “Know your tools”, I argue, is a central lesson of The Phaedrus – pointing us to early critiques of AI by Hubert Dreyfus (1972). Dreyfus first shows how “the story of AI” begins in yet another Platonic dialogue, The Euthyphro, which grounds the rule-based approaches of classic AI. These approaches, however, run aground on what becomes known as the frame problem – a problem, Dreyfus further shows, that can only be resolved by Socratic and Aristotelian forms of judgment / phronēsis. Dreyfus and Joseph Weizenbaum (1976) finally show that phronēsis is not computationally tractable.
For all their remarkable progress, then, contemporary AI, including LLMs, likewise fail to resolve the frame problem: their outputs are hence only semblances of judgment, thus presenting us with the same risks as writing for young Phaedrus.