Hands-On Learning in the Age of AI: Why Workshops Still Matter

Summary

Graham Dumpleton reflects on a PyCon AU 2026 talk about whether developer workshops remain relevant as AI assistants become better at answering technical questions and explaining setup procedures. He argues that passive walkthroughs are increasingly weak compared with an AI that can tailor an answer to a user’s release, operating system, and project. However, doing the work creates forms of learning that reading and chat do not reliably provide: mistakes have consequences, troubleshooting develops understanding, and a structured exercise can expose problems the learner would never think to ask about. Dumpleton says workshops commonly lose participants because of environment failures, steps that are too large, instructions that explain what to type without why, and missing verification. Hosted browser environments can remove setup friction, standardize participant state, and support deliberately broken scenarios or preconfigured multi-service systems for diagnosis practice. AI can help authors draft exercises, generate variations, handle boilerplate, and check old commands, but it tends to optimize for correctness rather than teachability. It cannot reliably judge step size, create productive uncertainty, or add checks unless the author explicitly guides it, and the result still needs testing with a real learner. The article also argues that workshop demand depends on how participants arrive: commercially motivated training and demos may retain their appeal, while voluntary self-serve learning competes directly with free AI answers and video. Dumpleton has released 24 free wrapture workshops running in JupyterLab on mybinder. Because wrapture is new and absent from model training data, the workshops can currently offer a form of practical help that generic AI cannot provide without being given the documentation. He presents the project both as a learning resource and as an experiment in using AI to create teachable hands-on material, while acknowledging that its effectiveness can only be established by watching people use it.