Anthropic's Claude Code event reveals coding's AI-driven future

At Anthropic’s Code with Claude event in London (coinciding with Google’s I/O), a packed room of developers showed just how far AI coding has come. During a talk by engineer Jeremy Hadfield, nearly half the audience raised hands when asked if they’d shipped a pull request written entirely by Claude in the past week. Even more striking: most of those same hands stayed up when asked if they hadn’t read that code at all. Anthropic claims that Claude now writes most of the code inside its own tools, including Claude Code. The company’s push for automation goes beyond simple code generation—it now aims to have Claude check and correct its own work through a feature called 'dreaming,' where agents write notes to each other and learn from past mistakes. Boris Cherny, head of Claude Code, said the new default should be 'having Claude prompt itself,' not humans prompting Claude.

Success stories from Spotify, Delivery Hero, and several vibe-coding startups painted a picture of teams fully trusting AI-generated code. Yet outside the conference halls, a growing backlash simmers. On Hacker News and Reddit, developers complain that AI-generated code adds review burden and erodes their own coding skills. 'The only people I've heard saying that generated code is fine are those who don’t read it,' one user noted. The tension highlights a critical question: as AI coding tools accelerate output, who ensures quality—and at what cost to human expertise?