# Ideas Worth a Longer Conversation: The JetBrains Research Podcast

[Jetbrains](https://yomu.fyi/company/jetbrains) · Katie Fraser · Aug 25, 2026

**Type:** Explainer

## Summary

The JetBrains Research Podcast investigates foundational software development and computer science questions through discussions with industry researchers. In team psychology, Cat Hicks demonstrates that collaborative culture and organizational belonging reduce developer anxiety around automated tooling while curbing the risks of overproduction pressure. Historical analysis by Tomáš Petříček shows that programming spans five competing cultures rather than pure specification implementation, explaining why automated code generation historically encounters limits. Regarding artificial intelligence evaluation, Ibragim Badertdinov details how SWE-rebench isolates model capability through dataset decontamination, leveraging the strict verifiability of software test execution for training feedback. Additionally, conversations with Alexander Kulikov and Anna Kogan examine how foundational mathematical understanding aids AI oversight and address adoption hurdles for open-source computer vision infrastructure.

## Context

Most public discussions surrounding artificial intelligence in software development focus narrowly on rapid productivity metrics, job displacement, and standard benchmark scores without examining fundamental research questions.

## Approach / What changed

The JetBrains Research Podcast interviews researchers across psychology, philosophy, algorithm design, coding agent evaluation, and computer vision to examine software engineering dynamics and machine learning practices.

## Takeaways

- Cat Hicks found that team signals of belonging, learning, and recognition cut developers' measured AI identity threat roughly in half, whereas high overproduction pressure causes developers to generate code without understanding it.
- SWE-rebench prevents training data contamination by evaluating coding agents exclusively on GitHub tasks created after a model's release date, revealing lower real-world problem-solving performance compared to SWE-bench Verified.
- Despite running on billions of devices and receiving 32 million monthly PyPI downloads, the open-source OpenCV library has historically been maintained by only two to eight developers at any given time.

**Tags:** [Developer Experience](https://yomu.fyi/topic/developer-experience), [LLMs](https://yomu.fyi/topic/llm), [Machine Learning](https://yomu.fyi/topic/machine-learning), [Open Source](https://yomu.fyi/topic/open-source)

[Read original post](https://blog.jetbrains.com/research/2026/08/jbr-podcast)