By Teresa Torres · producttalk.org · @ttorres on X · LinkedIn
Teresa Torres and Petra Wille argue that confident predictions about AI-driven change are often wrong, and that betting everything on one forecast is risky. Humans are poor at forecasting, and early adopters' experience rarely generalizes to everyone. Instead of trying to be right, they propose scenario planning: sketch several plausible futures, including extreme ones, and extract the underlying insights that should guide decisions today. The episode matters for product people navigating AI uncertainty because it offers a way to stay grounded without ignoring what's coming. Only the description and outline were available, so this metadata reflects the episode's stated framing rather than its full discussion.
01Key takeaways
- Treat any single forecast about AI as one possible future, not the future.
- Ask what else could happen before committing resources to a single bet.
- Don't assume early adopters' experience reflects how everyone else will behave.
- Run quick scenario exercises with your team to explore implications.
- Push ideas to extremes to reveal hidden assumptions and consequences.
- Extract the underlying pattern from a prediction rather than fixating on its specifics.
02Key sections
- The problem with predictions
- Headlines often claim certainty about where AI is heading, but the hosts question how reliable such forecasts really are. They note that strong predictions tend to ignore real uncertainty.
- Why experts get it wrong
- The discussion explores why even informed people misjudge the future, pointing to overconfidence and the tendency to project personal experience onto everyone.
- Scenario planning explained
- Rather than committing to one outcome, the hosts recommend mapping multiple possible futures and asking what else could happen.
- Early adopters versus reality
- Early adopters' enthusiasm is not representative of the broader market, so over-indexing on them can distort product decisions.
- Using scenarios in product work
- The hosts suggest quick team exercises, pushing ideas to extremes, and distilling the underlying insight rather than the exact prediction.
03From the post
“Listen to this episode on: Spotify | Apple Podcasts AI headlines are everywhere—and many claim they know exactly what’s coming next. In this episode, Teresa Torres and Petra Wille push back on that certainty. They explain why people are bad at predicting the future and why betting on a”
04Frameworks mentioned
Summary and takeaways written by PM Atlas; quotes are short excerpts. © the original author.