How Should I Be Using A.I. Right Now?

April 2, 2024

Guests: Ethan Mollick

Ethan Mollick is a professor at the Wharton School of the University of Pennsylvania who studies innovation and entrepreneurship; he writes the newsletter "One Useful Thing" about practical AI use and is the author of "Co-Intelligence."

Listen to the episode

About this episode

The first of three AI shows is a practical guide: how should ordinary people use AI right now? Ethan Mollick, Wharton professor and author of the "One Useful Thing" newsletter, tells Klein to spend ten hours experimenting with a frontier model and to "bring AI to every table." Treat AI less like a tool and more like a person — teachers and managers use it better than programmers — because it responds to feedback, not commands. Models have distinct "personalities"; giving the AI a persona is a technical trick that steers its probability space. Core techniques: chain-of-thought prompting and few-shot examples. On trust: hallucinations are dropping but never vanish — in one experiment, BCG consultants using AI did worse on tasks where it was confidently wrong. AI excels as a thought partner and creativity amplifier, not an oracle: 80th-percentile performance, a supplement for weakness rather than a replacement for expertise.

Read the transcript

Books recommended

0

No book recommendations recorded for this episode.