Why GitHub’s CEO Says Developers Must Embrace AI or Face Obsolescence

GitHub CEO Thomas Dohmke warned developers to either embrace AI or risk becoming obsolete, citing a new study and his own blog post.

He became CEO in November 2022 after serving as chief product officer, where he led AI tools such as GitHub Copilot.

Why the warning?

Interviews with 22 developers who heavily use AI tools show that AI is already a core part of their work, moving from skepticism to collaboration.

Four stages of AI adoption

Stage 1: AI Skeptic – developers try AI on small tasks, focus on code completion, and have low tolerance for errors.

Stage 2: AI Explorer – developers use AI for debugging, scaffolding, and code snippets, learning its limits.

Stage 3: AI Collaborator – developers actively co‑create with AI and develop prompt‑engineering intuition.

Stage 4: AI Strategist – developers treat AI as a powerful partner for feature development, complex tasks, and large‑scale refactoring, delegating implementation to AI agents while focusing on prompt design, verification, and validation.

New skills needed

AI fluency – understanding different AI tools and models.

Authorization and agent coordination – laying the groundwork for AI success.

Human‑AI collaboration – real‑time interaction with AI.

Fundamental programming knowledge – deep coding expertise remains essential.

Verification and quality control – rigorously checking AI‑generated work.

Product understanding – system‑level thinking beyond code.

Architecture and system design – guiding AI integration.

Education must change

Traditional computer‑science curricula that focus on memorising syntax are outdated; students now need to learn how to guide AI, critique its output, and think interdisciplinary.

Job outlook

The U.S. Bureau of Labor Statistics projects an 18% growth in software‑developer jobs over the next decade, but the work will shift from writing code to managing AI agents and validating their implementations.

Companies that boost developer productivity tenfold by leveraging AI can achieve exponential gains, and the most successful developers will be those who master AI‑driven workflows.

Not everyone will adapt

Some developers resist the transition, fearing that managing agents is unsatisfying, yet developers have always used programming languages to achieve results; the shift is real and ongoing.

Conclusion

Dohmke emphasizes that the transformation is already happening: early adopters are reshaping their roles, practices, and mindsets, preparing for a fundamentally different but more ambitious future of software development.