The Dangerous Illusion of AI Coding [video]

Great level-headed discussion on the erosion of knowledge and skill in organizations using AI. Which is an opposite direction to the hype about the promised super intelligence.

Howard makes a great analogy of LLMs interpolating between knowledge. This means LLMs can do combinatorial creativity but struggle with transformative creativity.

Also explains why LLMs are good at coding but not software engineering at a high level.