It AI n't What You Think!
summary
Argues that modern AI behaves as AI as accelerated inference: it repeatedly proposes plausible solutions but hallucinates facts and is arbitrarily inconsistent (example: United chat warning, Air Canada refund case). Demonstrates Claude 3.7 and LLMs reliably AI nails defects like parallel‑stream mutation and undefined ++ expressions but generates lower‑quality functional Java code. Proposes treating LLMs as investigative idea engines—use AI for ideas not solutions—and urges skill growth, critical thinking, and treating AI as a tool, not a substitute for responsibility.
AI-written summary · about
chapters (12)
- 0:00Introduction and curiosity
- 3:22AI as fascination and fear
- 7:43Revolution vs evolution of AI
- 12:18Transition periods in history
- 19:27AI accuracy and verification
- 25:11AI as a code investigator
- 29:28AI for code generation
- 33:21AI as accelerated inference
- 36:34AI as a tool and its limitations
- 42:25AI as a black swan event
- 47:22AI as the compiler of the future
- 53:51Adapting to change and critical thinking
related talks
56mAI: It AI-n't What You Think!
01h 01mKeynote: It AI-n't What You Think!
58mIt AI-n't What You Think!
01h 00mAI: It AI-n't what you think!
07mAI's impact on programming jobs and developer skills
25mAI vs. AI: How to Test AI Systems Using AI
21mAI's impact on software architecture
05mAI is your children, what is our future with them?