THE CRUNCH
GitHub's blog has published career advice arguing that AI is changing what it takes to stand out as a developer. Writing code still matters, the post says, but increasingly valuable skills include directing AI, evaluating its output, communicating tradeoffs and making sound technical decisions. Its first tip is to learn to direct AI rather than merely use it: as AI agents take on more implementation work, great execution means defining the problem clearly, providing the right context and deciding what is ready to ship. The post sketches a workflow where one developer coordinates several agents, one producing an authentication flow, another a documentation draft and a third a test suite, while remaining responsible for the outcome.
The post's second tip is to distrust AI's first answer. It suggests asking a second model to critique the first model's work, then applying your own judgment, noting that different models have different strengths and blind spots. As an example, GitHub points to its Copilot Rubber Duck agent, which uses a second model to critique plans, code and tests before you move on.
The third tip is to redirect the time AI saves towards bigger problems: understanding customer needs, weighing architectural tradeoffs and defining success metrics. GitHub's argument is that as agents take on more implementation work, judgment, communication and deciding what is ready to ship become the skills that actually distinguish engineers.