AI First Learning
summary
A concrete AI‑First system maps learning into repeatable workflows for planning, practicing, verifying and retaining skills, with guardrails for correctness and privacy; it introduces the GRAPES axes—grounded, reproducible, personalized, episodic—and shows failures (LLM hallucinations on ZFS flags, non‑deterministic answers and images) and wins (audience solves a seating logic puzzle with 12 vs the LLM’s 9). A VS Code extension ships curated prompts/responses to make examples deterministic, save tokens, and a companion AI‑first programming book ships summer 2026.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 2:02Great teacher traits
- 7:00Vikings and JFokus speakers
- 10:56LLMs fail as teachers
- 13:00Hallucinations in AI
- 17:01Non-deterministic AI responses
- 20:10AI’s coding inconsistency
- 24:19AI struggles to teach coding
- 27:28Books return in AI age
- 28:32LLMs lack personalization
- 31:20Word vectors and AI memory
- 33:05AI’s short-term memory
- 36:11Employers demand AI skills
- 37:13Schools lag in AI education
- 38:40AI first learning method
- 39:01AI first programming demo
- 40:50AI first programming book
- 41:56Conclusion & Q&A
related talks
05mAI is your children, what is our future with them?
26mThe present and future of programming with AI (Keynote)
55mAI-First Testing
25mAI vs. AI: How to Test AI Systems Using AI
59mIt AI n't What You Think!
32mComputer Programming Is Dead; Long Live AI-First Programming
36mEmbracing AI in Education: A Path to Inclusive and Effe...
09mLessons Learned in AI Engineering