@stevekrouse:

I'd like to add the benefits of "strolling" (any better word?) meaning to look around into different corners to learn and discover things. I really want to push back on the focus on maximum productivity.. when tinkering / strolling is fun and beneficial. Sure you can get the LLM to order pizza and have a full stomach in half an hour.. but Isn't life about taking unexpected detours, try new things and enjoying the path and people to get there? Now that I think of it, I actually use LLMs often for strolling into areas I'm not very familiar.. sort of a talk with a senior at the water cooler.

This talk clarified so many ideas around vibe & agentic coding for me. I must have watched it three or four times. I’ve spent about 30 hours with Claude Code on a few side projects recently. I couldn’t put my finger on why I was insisting to stay in this tight loop with the agent, critiquing every little change instead of giving it free range, until Steve defined “vibe coding” and connected it to the idea of legacy code. It resolved my cognitive dissonance to realize that vibe coding was just one way to use an agent, and that my way wasn’t just an intermediate step to get there. The way I’ve been using AI, I am using an agent, but I am writing the code. I’m not delegating to a teammate—I’m typing 200 characters to get external verification of my thinking and a first draft. I’m basically using conversational autocomplete.

Its no small feat challenging the human mind reflected through technology and society - hereby represented as "AI". Steve does a good job at relativizing the recent AI hype.