Fireside Chat, Interview, Panel
Who's Coding Now? - AI and the Future of Software Development
- AI is predicted to evolve into a higher-level language abstraction and potentially rewrite compiler design, with models serving as sparring partners for clarifying thinking and fetching real-world data via tools like "firecrawl."
- The coding market is expected to remain the second-largest AI sector behind consumer chatbots, potentially becoming the first truly large AI market, driven by its verifiable nature and developer motivation to build.
- Productivity gains from AI adoption could unlock approximately $3 trillion in value, which may recuperate the current $200 billion in annual AI investments.
- Future workflows may shift developers toward roles involving specifications or QA, while non-developers ("vibe coders") and enterprise users are expected to perform increasingly sophisticated tasks, potentially delaminating output and reasoning layers.
- Technical evolution includes a refocusing of CS education within five years toward problem statements and architecture, while niche manual optimization remains specialized; Python and formal languages are not expected to disappear entirely due to the need for precise scaling control.
- Legacy system migration faces specific hurdles with hardware-specifics and lost context, though generating specifications from legacy code before re-implementation is projected to yield compact, improved results.
- Risks and constraints include the potential unsolvability of certain distributed AI design problems, the difficulty of capturing runtime snapshots for migration, and enterprise adjustments to metrics acknowledging that AI cannot fully rule out undesirable outputs.