Nobel Prize-winning physicist Saul Perlmutter addressed artificial intelligence as a two-edged tool during a December 2025 interview on the "In Good Company" podcast, briefly comparing it to earlier concerns about calculators in education. Perlmutter, who won the Nobel Prize in Physics for discovering the universe's accelerating expansion, discussed how AI intersects with the critical thinking methods he teaches.
The physicist noted that AI can give students the impression they have actually learned the basics before they really have, potentially leading them to rely on it too soon before they know how to do the work themselves. He identified a particular concern with the current generation of AI being very good at being overly confident about what it's saying, which users may accept without scrutiny because it's typed on the screen.
Perlmutter teaches a critical thinking course covering 24 concepts and has asked students to think a bit hard about how to use AI to make it easier to operationalize each concept in their day-to-day lives, and also how to use these concepts to tell whether AI is fooling them or sending them in the right or wrong direction.
The physicist noted that when users know these different tools and approaches to thinking about problems, AI can often help them find the bit of information they need to use these techniques.
Notes: This post was drafted with the assistance of AI tools and reviewed, fact-checked, edited, and published by humans. Image: DIW-Aigen
Read next: AI Video Translation Offers Efficiency Potential but Human Nuance Remains Key
The physicist noted that AI can give students the impression they have actually learned the basics before they really have, potentially leading them to rely on it too soon before they know how to do the work themselves. He identified a particular concern with the current generation of AI being very good at being overly confident about what it's saying, which users may accept without scrutiny because it's typed on the screen.
Perlmutter teaches a critical thinking course covering 24 concepts and has asked students to think a bit hard about how to use AI to make it easier to operationalize each concept in their day-to-day lives, and also how to use these concepts to tell whether AI is fooling them or sending them in the right or wrong direction.
The physicist noted that when users know these different tools and approaches to thinking about problems, AI can often help them find the bit of information they need to use these techniques.
Notes: This post was drafted with the assistance of AI tools and reviewed, fact-checked, edited, and published by humans. Image: DIW-Aigen
Read next: AI Video Translation Offers Efficiency Potential but Human Nuance Remains Key