Lecture IX — Judgment and Presenting
AI-Assisted Programming for PhD Researchers
When Not to Use AI
The Question Is Not “Can It?”
- It usually can produce something — that was never the question
- The question is whether you should delegate this
- Now, in a thesis you will have to defend
- “It compiles” and “I can stand behind it” are different bars
Do Not Delegate
- What you are here to learn — the core methods of your field
- Architectural decisions you will live with for years
- Anything you cannot verify yourself
- Statistics you do not actually understand
The Verification Boundary
- Usable delegation ends where your ability to check ends
- Self-perception is a bad meter for this
. . .
- METR’s developers felt ~20% faster while being 19% slower (Becker et al. 2025)
- Growing that boundary outward is your development as a researcher — the goal is calibrated trust
Keeping Your Edge
- Deliberately code without AI sometimes — small katas, first drafts of core logic
- Make explain-back a habit: say what the code does before you accept it
- Confidence in the tool predicts less critical thinking (Lee et al. 2025)
- The RCT, one last time: how you use it decides what you keep (Shen and Tamkin 2026)
Your Own Guardrails
- When the goal is learning, ask the agent for hints and next steps, not full solutions
- That is the exact design that erased the learning harm in the PNAS tutoring RCT (Bastani et al. 2025)
- The effect is specific to that guarded-tutor study — treat it as a pattern to copy, not a guarantee
- After this course, you are your own guardrail
Discussion
Two questions for the room:
- Which task this week would you not delegate again?
- Where did you accept something you couldn’t explain?
Presenting Your Project
The Shape of a Good 12 Minutes
- Problem — what and why (2 min)
- Live demo — show it working (4 min)
- Process — how you worked with the agent (3 min)
- Challenges — what broke, what you changed (2 min)
- Insights — what the room should take away (1 min)
Mirrors the four points you are graded on.
Demo Risk Management
Live demos fail. Have a fallback:
- Record a 60-second screen capture now, while it works.
- Test on the projector’s resolution before you present.
Slides, Fast
- You may of course use AI for the slides too
- Same rules apply: every claim on a slide is yours
- Keep it lean — aim for ~8 slides
- Talk to the demo, not to a wall of text
Reflection Beats Perfection
- An honest “here is where it went wrong and what I changed”…
- …scores better than a polished façade
- It also teaches the room more than a flawless run
- Your judgment is the thing on display, not the tool’s output
Continue Your Journey
After the Course
- Keep the three habits: the loop, the checks, the review
- This website stays up — come back to it
- Go deeper: course literature and references
- Details for the final talk: Presenting Your Project
- Questions later? tobiasvlcek.com
References
Bastani, Hamsa, Osbert Bastani, Alp Sungu, Haosen Ge, Özge Kabakcı, and Rei Mariman. 2025. “Generative AI Without Guardrails Can Harm Learning: Evidence from High School Mathematics.” Proceedings of the National Academy of Sciences 122 (26): e2422633122. https://doi.org/10.1073/pnas.2422633122.
Becker, Joel, Nate Rush, Elizabeth Barnes, and David Rein. 2025. Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity. arXiv. https://doi.org/10.48550/ARXIV.2507.09089.
Lee, Hao-Ping (Hank), Advait Sarkar, Lev Tankelevitch, et al. 2025. “The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects From a Survey of Knowledge Workers.” Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems (Yokohama Japan), April, 1–22. https://doi.org/10.1145/3706598.3713778.
Shen, Judy Hanwen, and Alex Tamkin. 2026. How AI Impacts Skill Formation. arXiv. https://doi.org/10.48550/ARXIV.2601.20245.