Machines, Learning, and Machine Learning
summary
An Omron X4 blood‑pressure cuff shows how humans project intelligence onto devices while large language models like ChatGPT, Claude and GitHub Copilot reshape development workflows. Copilot generates a Ruby Jackal (Jekyll) plugin that first runs, then introduces a commit that breaks the build despite an instruction to snapshot. Demos produce a purple .NET 8 login system and 5,000‑line PRs, revealing variable ratio reinforcement, hallucinations, non‑determinism, and sudden token pricing that make delivery, review and legal accountability the real problems.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 2:18Projecting intelligence onto machines
- 11:49Determinism in software
- 18:07Non-determinism in real systems
- 24:09AI bubble economics
- 30:26AI and reinforcement in dev
- 36:02Delivery vs development
- 41:15Human accountability
- 47:12Career advice
- 57:15Local AI development
- 1:01:27AI limits and Douglas Adams
related talks
50mMachines, Learning, and Machine Learning
01h 03mIt's All About AI
32mTesting Machine Learning Systems: What it is and why you should care
59mMaking the most of artificial and human intelligence for data science
54mMachine Learning for Developers
55mEveryday AI: Practical Applications Beyond the Chatbot
40mMachines are Learning
46mMultimodel Multimodal and Multiagent innovation with Azure AI