Your AI Needs an Assistant
summary
Argues that generative models are great at ideation but must be paired with verification tooling so code actually works, and demos an iterative pipeline: prompt Gemini for a TypeScript function, ask for unit tests, run tests and lint and typecheck, feed failures back to the model, repeat until tests and TypeScript types pass, using tools like TypeChat, LangChain, Sourcegraph/Cody and ESLint/Mocha. Shows the loop fixes real bugs (null vs undefined, implicit any) and ends with a working, type safe implementation.
AI-written summary · about
chapters (15)
- 0:00Introduction and speaker intro
- 1:49AI introduction and memes
- 3:37AI potential and pitfalls
- 5:45Gartner hype cycle explained
- 7:30Five ways to use AI effectively
- 8:13Read the documentation first
- 10:21Static analysis benefits
- 13:52Low trust code and linting
- 14:33AI for unit tests
- 15:40Combining dynamic and static analysis
- 17:27TypeChat for data validation
- 19:57AI strengths and verification
- 21:19Demo with TypeChat and AI
- 23:06Predictions for AI future
- 24:10Recap and closing
related talks
26mAI and the Future of Testing
45mPanel Discussion on AI for Code Generation
42mBuild Without Limits - Lessons Learned Working with Data and AI Coding Assistants
47mTest-driven Development, AI, Pair Programming, Agile Testing
39mBuild without limits: lessons learned from using generative AI dev tool
01h 26mAutomation of the Automation using GenAI
58mAI, Determinism, Code Generation and Refactoring
26mAl and Software: The Good, The Bad, The Scary