Talk Python To Me #532: 2025 Python Year in Review
41 snips
Dec 29, 2025 Guest
Barry Warsaw
Guest
Reuven Lerner
Guest
Brett Cannon
Guest
Jodie Burchell
Guest
Thomas Wouters
Guest
Gregory Kapfhammer
Join a powerhouse panel featuring Gregory Kapfhammer, an expert in software engineering, Thomas Wouters, a key figure in Python's core development, Jodie Burchell, a data scientist at JetBrains, Brett Cannon, a seasoned Python developer at Microsoft, Reuven Lerner, a Python trainer, and Barry Warsaw, a veteran core contributor. They discuss the future of Python in 2025, exploring AI's evolving role, the impact of lazy imports and free-threading, and the need for improved community funding. Get insights on user experience enhancements and the balance between simplicity and debugging challenges.
AI Snips
Chapters
Transcript
Episode notes
AI Hype Meets Practical Limits
- Large LLM hype met reality in 2025 as improvements hit diminishing returns and specialized models rose.
- Jodie Burchell warns the AGI narrative outpaced measurable progress and expectations should be tempered.
Give LLMs Fast Feedback Loops
- Teach LLMs to use fast tooling that gives rapid feedback; that boosts their ability to iteratively refactor and add tests.
- Design agents to call quick validators and run small checks rather than relying solely on model output correctness.
Using LLMs As A Brainstorming Partner
- Reuven Lerner uses LLMs as a sparring partner for writing and editing newsletters rather than full code generation.
- He finds model feedback dramatically improves quality even without auto-generating code.