AI is changing developer work. Here are three skills to strengthen.

A GitHub Blog post titled "AI is changing developer work. Here are three skills to strengthen" describes how developer workflows are shifting as AI tools become more capable and more integrated into day-to-day work.

The first skill is learning to direct AI rather than just use it. The post says great execution increasingly means defining the problem clearly, providing the right context, evaluating AI-generated code, and deciding what is ready to ship. It contrasts a traditional authentication workflow — create branch, write code, run tests, open pull request — with a workspace in which multiple agents produce an authentication implementation, a documentation draft, and a test suite. The post states the developer remains responsible for the outcome while spending less time implementing each piece.

The second skill is not trusting AI's first answer. The post says the first answer is not always the best, and that experience writing clean, maintainable code helps evaluate AI output. It suggests asking a second AI model to critique the first model's work, then using judgment to evaluate both responses. An example shows one model generating a SQL query for each customer's most recent order and a second model flagging that it does not handle duplicate timestamps, is missing an index recommendation, and may perform poorly on large tables. The post says GitHub Copilot's built-in Rubber Duck agent uses a second model to critique plans, code, and tests.

The third skill is using AI to solve bigger problems. The post lists work developers can take on when implementation time is saved: understanding customer needs, evaluating tradeoffs, designing better systems, and making technical decisions. An example issue for adding dark mode shows AI producing implementation, tests, and documentation, alongside a developer checklist covering validating the customer problem, reviewing architectural tradeoffs, checking accessibility, defining success metrics, and approving the solution.

The post concludes that learning to work effectively with AI and strengthening technical judgment can help developers adapt. It appeared on The GitHub Blog.