Lecture XII - Presentations and Discussion
Programming with Python
EP 12: Demo Day
The investor is in the room
Back in Session VI she stood at a whiteboard and asked why everything here was built from scratch. Today the answer is on your screens: built by you, with the standard library, pandas, plots, git, and an agent you learned to supervise.
The rules: 7 minutes presenting, 3 minutes of questions, order drawn at random.
While you watch
- Note one thing you would have done differently, and one thing you are taking for your own next project
- Questions are part of the format: specific, kind, useful
Your Future with AI
AI and your first job
- US: in jobs AI can largely do, employment of 22-25-year-olds trails their peers by 19%, through fewer hires, not layoffs (Brynjolfsson et al. 2026)
- Sweden: a smaller gap of 5.5%. Job ads fell with interest rates rather than with AI (Lodefalk et al. 2026)
- Denmark: accountants, HR, marketing, office clerks: no effect on pay or hours above 2% (Humlum and Vestergaard 2026)
So far: fewer junior hires, no economy-wide job losses, size disputed. All three are working papers.
What AI does to work and skill
- BCG consultants: with AI, 12.2% more tasks, 25.1% faster. On a task beyond AI’s reach, 19% less likely to be right (Dell’Acqua, McFowland, et al. 2026)
- Patent lawyers: AI lifted the juniors’ work, but judgment without AI grew only for the seniors (working paper) (Autor et al. 2026)
- Developers learning a new library: 17% lower on the quiz with AI, worst on debugging (preprint) (Shen and Tamkin 2026)
- At school, an AI tutor giving hints, not answers, largely removed the learning loss (Bastani et al. 2025). Part I worked the same way
What stays valuable
- At P&G, one person with AI matched a team without it (Dell’Acqua, Ayoubi, et al. 2026)
- Today you showed projects built with an agent you supervised, twelve sessions after your first
print() - The verification boundary: usable delegation ends where your ability to check ends. Growing that boundary outward is your career
- Your edge, deliberately kept: explain every line before you accept it, and work without AI sometimes
Discussion: from junior to senior
One minute with your project group, then three voices from the room:
If AI drafts most junior work, how does a junior analyst become a senior one?
Send-off
Keep programming
The skill fades if it sits idle. A few ways to keep it alive:
- Use Python in your thesis: data cleaning, analysis, plots you can defend
- Find a way to apply it at work: the repetitive task nobody wants to do by hand
- Advent of Code: free, ad-free programming puzzles, one a day from December 1st. A genuinely fun way to keep the muscle warm.
Thank you
- We covered the basics of Python: from a first
printto a real project on your own machine - I hope you enjoyed it, and I hope it’s useful long after the grade
- If you have questions or feedback, please tell me
All the best for your studies and your career. Now go build the thing.
Literature
Keep reading
The literature list of this course has more books, several of them free online.