Machines, Learning, and Machine Learning
summary
A mix of concrete failures and experiments: a smart blood‑pressure cuff reports health scores 5→10 because it reads calendar dates; a haunted test traces to a 15‑minute sliding cache expiration that breaks for Nepal’s UTC+5:45 offset; Copilot/Claude quickly generates a working web login + PostgreSQL and a Jekyll plugin that sometimes deletes files, so produced HTML is usable even when the plugin code is opaque. Practical takeaways: run local models (LM Studio), learn HTML/CSS/JS and robust delivery concerns, and beware vibe coding and variable ratio reinforcement.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 3:02Why technology matters to our lives
- 9:18Determinism vs reality
- 14:34AI rise and limits
- 20:13AI tools and non-determinism
- 25:48Programs vs products
- 33:48AI replacing engineers?
- 38:02Advice for developers
- 45:02AI and human creativity
- 48:10Closing thoughts
related talks
01h 04mMachines, Learning, and Machine Learning
32mReliable Machine Learning
38mBuilding robust machine learning systems
32mTesting Machine Learning Systems: What it is and why you should care
24mChallenges in building Machine Learning models in production
25mUsing Machine Learning To Improve UI/UX
18mReliable Maintenance of Machine Learning Models
42mLearning Machine Learning