The biggest findings on AI agents, top languages and fastest-growing developer communities

The languages gaining the most ground are the ones where developers rely on AI the most. TypeScript, Python and Rust are growing faster than any others, and AI had a hand in roughly 40% to 44% of the pull requests merged in them, more than in any other language, according to GitHub. In the languages developers are moving to fastest, AI help is part of a large portion of the finished work. That raises a harder question than which language is popular: whether developers bring AI into these languages, or whether AI leads developers to choose them.

The influence probably runs in both directions. Developers get better results from AI tools in languages with strict rules and clearly declared types, Andrea Griffiths, a senior developer advocate at GitHub, tells The New Stack, and that makes such languages more appealing. GitHub sees a sign in its data that people who code with agents may favor languages in which reviewing an agent's code is simpler. If so, typed languages may gain twice, because AI writes them more reliably and people can catch its mistakes more quickly. For a developer deciding what to learn next, how well AI handles a language has become part of the reason to choose it.

Ease of checking does not explain every gain, because AI also makes tedious work easier to bear. Shell $SHEL scripting, a famously fiddly kind of code with little in common with strictly typed languages, grew 206% in AI-generated projects, according to GitHub data reported by The New Stack. Developers may use more shell code because AI removes the drudgery, not because they have come to enjoy it. Fast growth in a language is therefore a weak signal of what developers actually prefer.