InsightfulDiscussion

How to solve hard problems - mathematicians explain | Lex Fridman

Lex Clips

A mathematician discusses their problem-solving methodology, emphasizing the importance of building mental models, learning from neighboring fields, and maintaining long-term research vision. They advocate for collaborative exploration with students on problems adjacent to their expertise rather than purely within it, and highlight the value of peripheral observation alongside focused deep work.

Summary

The speaker describes their approach to tackling hard problems through the development of mental world models that inform how they understand results across sub-areas. A key evolution in their advising philosophy involved shifting from giving students problems entirely within their own expertise (where quick dismissal of ideas prevented productive collaboration) to assigning problems 10-20 degrees removed from their area of specialization. This approach creates a collaborative learning environment where both advisor and student explore unfamiliar territory together, reverse-engineering why results work rather than mechanically parsing papers line-by-line. The speaker emphasizes that discovering key ideas that shift perspective often leads to identification of new approachable questions through the unique organizational framework developed during this joint exploration.

The speaker reflects on how AI technology may facilitate this type of cross-disciplinary work while simultaneously raising concerns about losing the human element of expert explanation. They note a concerning trend among students who think in 3-6 month horizons due to anxiety about AI disruption, rather than the multi-year perspective necessary for meaningful research. The speaker advocates for working on "future-proof problems"—research directions that will remain valuable regardless of technological advances.

Regarding problem-solving methodology, the speaker describes a tension between laser-focused engineering work on singular problems and maintaining mental freedom for peripheral observation and curiosity. Solutions often emerge not from direct frontal attention but from peripheral vision and casual observation when turning over rocks in unrelated fields. The speaker gives an example of how fluid dynamics, biology, and material science form a fruitful neighboring cluster due to shared characteristics: multi-scale phenomena, non-linear dynamics, and messy, complex systems with nebulous objectives.

Key Insights

  • The speaker changed their advising approach from assigning problems entirely within their expertise (where they could quickly critique ideas) to assigning problems 10-20 degrees off their area, creating genuine collaborative learning where neither party knows all the answers.
  • When exploring unfamiliar results together, the key breakthrough comes from identifying one or more perspective-shifting ideas rather than understanding all details, after which new adjacent problems become approachable through the newly developed organizational framework.
  • Students are increasingly thinking in 3-6 month time horizons due to anxiety about AI disruption, whereas meaningful research requires multi-year vision; the speaker advocates only working on problems one will still care about years in the future.
  • Solutions to problems are often found through peripheral vision and casual observation of unrelated systems rather than through direct, focused frontal attention to the problem itself.
  • Biology, fluid dynamics, and material science form a productive neighboring cluster because they share common mathematical characteristics: multi-scale phenomena, non-linear dynamics, and complex systems with messy, nebulous objectives.

Topics

Problem-solving methodology and mental modelsCross-disciplinary and neighboring field learningAdvisor-student collaboration on researchLong-term vs. short-term research thinkingBalance between focused work and peripheral observationAI and future of expert knowledge transferMulti-scale complex systems in natural sciences

Transcript

[0:02] Actually, that makes me think, can you actually describe how your thinking process of how you think through really hard problems? Is there some aspects to it that are interesting, like a lot of coffee in the morning, at night, alone, brooding with a cigar or on a together at a whiteboard with others? Is there some interesting color, texture, insights you can give? - Sure. I think a lot through world models in [0:36] some way. So, you know, I have some mental model of how I think results in some sub-area go, and I think that colors a lot of how I see surrounding areas too. So for me, a lot of my work is really about,…

Full transcript available for MurmurCast members

Sign Up to Access

More from Lex Clips

Get AI summaries like this delivered to your inbox daily

Get AI summaries delivered to your inbox

MurmurCast summarizes your YouTube channels, podcasts, and newsletters into one daily email digest.