OpenAI and Anthropic’s new models spark debate over the future of traditional programming

The release of powerful new coding models from OpenAI and Anthropic has prompted developers to seriously question whether traditional programming is becoming obsolete.

Last week, the companies unveiled GPT-5.3-Codex and Claude Opus 4.6, both showcasing a major leap in AI-driven coding capabilities.

GPT-5.3-Codex demonstrated significantly improved performance in benchmark tests, while Opus 4.6 introduced the ability to deploy autonomous teams of AI agents to tackle different aspects of complex projects simultaneously. Both models can write, test, and debug code with minimal human involvement.

The launches triggered intense debate within the engineering community. A particularly high-profile contribution came from Matt Shumer, CEO of OthersideAI, who argued in an essay that AI models can now autonomously handle the entire development cycle — generating tens of thousands of lines of code, launching applications, testing features, and iterating until satisfactory results are achieved, while developers simply describe the desired outcome.

The piece drew mixed reactions. Reddit co-founder Alexis Ohanian endorsed the argument, while NYU professor Gary Marcus dismissed it as “weaponized hype,” criticizing the lack of evidence that AI can reliably build complex, error-free applications.

Many engineers say they have already stopped writing code line by line, instead orchestrating AI systems. Spotify co-CEO Gustav Söderström revealed that some of the company’s top developers “have not written a single line of code since December.” Spotify’s internal tooling uses Claude Code for remote deployment, allowing engineers to issue instructions to AI via Slack during their commute.

At Anthropic, between 70% and 90% of the company’s code is now generated by AI. OpenAI claims GPT-5.3-Codex is the first model that “was instrumental in creating itself.”

Still, some developers warn of burnout risks. Veteran engineer Steve Yegge described falling asleep after extended coding sessions and argued that AI tools can exhaust developers through repetitive cognitive patterns.