Coding Of The Future Will Be More About Reviewing AI Code Than Writing Code: Anthropic Researcher

Even if coders aren’t fully disrupted with the advent of AI, their roles might become radically different with the advent of AI.

The role of the software developer, long defined by the intricate art of writing code, is on the verge of a historic transformation. According to Boris Cherny, a researcher at AI company Anthropic, the future of programming will pivot from manual creation to high-level oversight. In this new paradigm, developers will act less as hands-on builders and more as expert reviewers and conductors, guiding AI agents that do the bulk of the coding.

Cherny, whose work places him at the cutting edge of AI model development, describes this as a necessary evolution in mindset for the entire industry. “There’s a bit of a mental shift that has to happen. Some people really love controlling the code,” he states. “If you’re used to handwriting code, the industry is now shifting to a place where you’re orchestrating agents that write your code. It’s more about reviewing code than handwriting code.”

While this transition may be jarring for some, Cherny frames it as a leap forward in productivity and capability. “People have to deal with this transition. As a programmer, it’s incredibly exciting because you can do so much more, so much faster,” he says. The efficiency of modern AI tools is already potent enough to change his personal preference. “There is still some stuff where I’ll have to dip down and hand-write code, but now I kind of dread it because AI is just so good at it.”

Looking forward, Cherny predicts that the need for manual coding will continue to shrink as AI models become more sophisticated and context-aware. “As models get more capable, these windows where you have to hand-write code—either because it’s a complex data model, something really sophisticated like the interaction between a bunch of system components, or something that’s hard to type out in a prompt—will keep receding,” he argues. “More and more, programming will be about orchestrating agents.”

Cherny’s vision of a developer as an “AI orchestrator” is rapidly becoming a reality. This marks a fundamental evolution of the required skillset, away from pure syntax and algorithm implementation towards critical thinking and architectural oversight. The most valuable developers will be those who can provide clear, high-level instructions to an AI, critically assess the output for security vulnerabilities, logical errors, and adherence to best practices, and effectively manage the integration of various AI-generated components. The job is shifting from building the bricks to designing the cathedral and ensuring its structural integrity.

This industry-wide change is being propelled by a new class of powerful tools. Applications like Cursor and Windsurf have become commonplace. More advanced, agentic systems are also emerging. Anthropic’s own “Claude Code” is designed as an agentic tool that lives in the terminal, understands an entire codebase, and can handle complex workflows from bug fixes to feature implementation through natural language commands. This trend is mirrored by other autonomous AI software engineering agents that aim to take on entire development projects from a simple prompt. For businesses, this promises an unprecedented acceleration in software delivery and innovation. For the developer, it signals a clear mandate: the future isn’t about writing more code, but about leveraging AI to think bigger.

Posted in AI