Why
Every language community claims to be ready for AI. I wanted a way to check that claim that would survive someone disagreeing with me. So I built a framework where every score traces back to a cited source, every claim can be countered, and the predictions are dated so they can be graded.
What it is
Ten languages, scored on eleven weighted dimensions: C++, C#/.NET, Elixir, Go, Java, Kotlin, Python, Rust, Swift and TypeScript. The heaviest weight goes to how well AI agents can operate the toolchain. Other dimensions cover how easy the language is for a machine to analyse, native versus ecosystem AI support, structured-output support, dependency risk at library level, and how plausibly a language can close its gaps.
Behind the scores sit about 390 atomic claims with primary-source citations and 173 counter-links. Validators run in CI and enforce the chain from source to claim to evaluation to insight. The framework is schema-locked at version 0.6, with an honest list of what version 1.0 still needs: multi-rater calibration and external review.
The thesis
The AI era favours languages that are easy for both humans and machines to analyse, verify, refactor and operate, and whose key dependencies are resilient enough to bet a five-year platform on. The ecosystem risk dimension turned out to matter more than I expected: a single-maintainer library at the heart of a language’s AI tooling is a strategic risk, not a detail.
Accountability
The ranking ships with falsifiable predictions. I will grade them in public in May 2027, because a framework is a bet, not an oracle.