Teaching and learning data science in the era of AI
summary
Demonstrates using GitHub Copilot to generate a horizontal ggplot from the religious income dataset and shows one prompt that produces a correct chart and a slight rewording that makes Copilot hallucinate nonexisting variables. Argues that missing domain terms like pivot_longer and missing vocabulary — not just prompt tuning — cause failures, and proposes pedagogy that gives students the bad prompt and output to debug the prompt and explain what's absent.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 0:45Example
- 1:25New learner reaction
- 1:55What went wrong
- 2:46Learning Pivot
- 3:20Updating pedagogy
related talks
36mEmbracing AI in Education: A Path to Inclusive and Effe...
42mSimplifying Data Analysis with GitHub Codespaces, Jupyter Notebooks & Open AI
22mAI, Make It So - A Dive into the Universe of Prompt Engineering
25mDemocratising learning of AI - experiences from the University of Oxford
32m"How Smart is AI? – Real projects to build real understanding"
45mTest Driven Prompt Programming: How AI is Changing the Way We Code
01h 02mPANEL - Unleashing the Future: The Impact of Gen AI on Development Productivity
59mMaking the most of artificial and human intelligence for data science