What's worth building when AI can build anything?

Categories: Product, Tools

Summary

When AI can build anything, the real competitive advantage isn't speed—it's understanding system complexity and ripple effects. Sue Black's research on software dependencies reveals that experienced developers who map interconnected systems are increasingly valuable, not obsolete, as AI handles routine coding tasks.

Key Takeaways

  1. Latent experience in understanding system dependencies and ripple effects across codebases is desperately needed even as AI agents improve; this knowledge cannot be easily replicated by models.
  2. The core challenge in large systems (20M+ lines of code in banking sector) isn't making individual code changes—it's predicting cascading effects across interconnected components when documentation is outdated.
  3. Industry moves too fast for documentation to stay current, creating a gap where human expertise becomes the map of how systems actually work, not how they're documented to work.
  4. The existential question for builders isn't whether AI can build code, but whether to build everything possible or be more mindful about what's worth building; this requires human judgment about system impact.
  5. Kent C Dodds shifted from feeling threatened by AI models doing his job 'better' to optimism by focusing on agent coordination and education—reframing human value toward orchestration rather than solo execution.

Related topics

Transcript Excerpt

Please welcome Sue Black and Kent C Dodds. Hello. Hi. Hi, everybody. We're so excited to talk with you. Before we get into anything, I want to ask a show of hands. How many of you are confident that what you are going to be doing a year from now is exactly what you're doing right now? Okay, good. I didn't want to embarrass anybody. Nobody raised your hand. There are all bets are off. I can't imagine what where we were a year ago and to be where we are now. It just couldn't have predicted that at all. Yeah. So thank you very much. So we're going to talk about what is worth building when AI can build anything. And to to get us into that, I think it'd be useful for us to get some context on who we are. So, Sue, why don't you introduce yourself to everybody here? Thanks very much, Ken. Hi, eve…

More from GitHub