Too Lazy to use Claude Code for Research (with YY Ahn) Artwork

Too Lazy to Read the Paper

In this podcast the author explains a paper to me, your host, Professor Sune Lehmann (https://sunelehmann.com). The participants are authors of a paper in network science or data science. Sometimes I feature a group of co-authors! The intended audience is PhD students, PostDocs and other scientists. The idea is to start with a bit about the paper's author, the idea for the paper. Then talk about the research itself. And we’ll end by gossiping about the reviewing process, etc. The whole thing is based on the idea that papers are so formal. And that when two people talk to each other informally, it’s often more fun – and tends to get ideas across more effectively.

All Episodes

Too Lazy to Read the Paper

Too Lazy to use Claude Code for Research (with YY Ahn)

June 02, 2026 • Sune Lehmann • Season 3 • Episode 1

0:00 | 1:16:36

In this video I talk to YY Ahn about using Claude Code for Research.

Normally in Too Lazy to Read the Paper, somebody explains a paper to me, because reading papers is hard and talking to people is fun and natural.

And if someone's too lazy to read the paper, they're definitely too lazy to read the manual.

So today there's a new format. There's no paper. Instead there's a thing we're all talking about in the shadows – agentic coding, Claude Code. The whole business of handing an AI your research and letting it write and run the code – and my sense is that getting started on this is a big hurdle to a lot of folks. So to help them (and myself, I'm sitting down with all-round genious, renessaince man, deep thinker, conneseur of all things coding and living in the terminal, and honest, decent person, YY Ahn.

YY is
* Quantitative Foundation Distinguished Professor
* School of Data Science
* University of Virginia

We cover the following topics

* Intro to YY
* Brief intro to Claude Code … what even is it?
* Working in the terminal
* Claude as reviewer zero
* Submitting to arXiv, checking references
* Some science use-cases
* The concrete setup … remote control, etc. What about context windows, should Claude be re-started once in a while? Do you have tasks that run on a schedule.
* What about Gemini or ChatGPT as an alternative. Having interchangble coding engines.
* Should one have a local model that the AI can direct.
* Cool tips & tricks for making life easier
* Discussion about what this does to science, where can humans still make a difference
*Discussion about de-skilling