Conversation, Computation, and Community: Solving Scientific Problems with Jupyter Notebooks and AI Tools

JetBrains PyTV Event: Python Unplugged on PyTV, March 4, 2026.

Transcript

1. Carol Willing | March 4, 2026 Conversation, Computation, and Community Solving Scienti fi c Problems with Jupyter Notebooks and AI Tools 1
3. PROBLEM 3
4. opportunity or overwhelm 4
5. PERSPECTIVE 5
6. My perspective • 60 years on this earth • 50 years programming for fun • 39 years using AI tools • 14 years coding with Python AI Tools 39 6
7. Superpowers, leadership, and learning Python and Science speakerdeck.com/willingc 7
8. Rule 1 AI tools serve humans 8
9. Rule 2 Humans drive AI tools 9
10. Naming things is hard. AI is not a person or an all-knowing wizard. Try to be speci fi c when you talk about AI tools. Rule 3 10
11. Learn creative problem solving Rule 4 success 11 failure
12. START 12
13. 1 Hypothesis 3 Results 4 Critical thinking 2 Experiment Problem Solving and the Scienti fi c Process 13
14. Donald E. Knuth, Literate Programming (1984) Instead of imagining that our main task is to instruct a computer what to do, let us concentrate rather on explaining to human beings what we want a computer to do. Literate Programming 14
15. 2015 15
16. Literate Problem Solving events.linuxfoundation.org/ai-in-jupyter Conversational Interactive and iterative Pragmatic and understandable Rigorous process and results Transparent and explainable 16 2026
17. HOW 17
18. Pillars of Python Problem Solving Conversation Computation Community 18
19. CONVERSATION 19
20. Conversation The heart of scienti fi c inquiry Bridges communication and computation Captures context Foster discoveries across scienti fi c domains Helps global research teams with AI translation tools Records process which reduces impact of interruptions Productivity tip Try speech to text tools. Speaking is often faster than typing. 20
21. COMPUTATION 21
22. Computation Work fl ows and processes Experiment with AI tools and software development practices. Protect against unexpected changes to data. Control the permissions that you give to agents. Remember agents work on your behalf. Be responsible. 22
23. Computation Active and iterative Focus more on ideas than tools. Measure what is important and create benchmarks. Try to break things. Think like a hacker. Challenge the process and the results. 23
24. COMMUNITY 24
25. Education and community building scienti fi c-python.org pyopensci.org Scienti fi c Python carpentries.org 25
26. Active participation Expertise Pragmatic Rigor Transparent Knowledge sharing 26
27. Pragmatic voices Simon Willison https://simonwillison.net/ Hamel Hussain https://hamel.dev/ Sebastain Raschka https://magazine.sebastianraschka.com/ Katherine Jarmul https://probablyprivate.com/ Cat Hicks https://www.drcathicks.com Eric Ma https://ericmjl.github.io/blog/ Marimo Team https://marimo.io Hugo Bowne-Anderson https://hugobowne.github.io/ Audrey Feldroy https://www.feldroy.com/ Hilary Mason, Hidden Door https://www.hiddendoor.co/about Samuel Colvin, Pydantic https://pydantic.dev/about plus many more talented folks in science and Python ecosystems My follows for insights on pragmatic problem solving 27
28. RESULTS 28
29. https://hms.harvard.edu/news/new-arti fi cial-intelligence-model-could-speed-rare-disease-diagnosis https://github.com/debbiemarkslab/popEVE 29
30. ACT 30
32. THANK YOU 32
33. Attributions 33 2, 31: Carol Willing 4: Giphy (links in GIF) 6: Napari and CZI 7: pt, Adafruit 8-11, 22, 23: Apple/OpenAI Generated 15, 16: Jupyter 25: pyOpenSci, Scienti fi c Python, Carpentries 29: Cited on page 32, 33: PyBeach logo
34. Carol Willing | March 4, 2026 Conversation, Computation, and Community Solving Scienti fi c Problems with Jupyter Notebooks and AI Tools 34