The End of the Polyglot Programmer Era: AI Knows All Languages Better Than Humans
Details of the new IEEE ranking and the reasons behind the unexpected decline in interest in web development.
Details of the new IEEE ranking and the reasons behind the unexpected decline in interest in web development.
Since 2013, IEEE has annually published an interactive ranking of the most popular programming languages. But today, the conventional ways of measuring popularity may be losing their meaning—all because of how the code-writing process itself has changed.
In the latest IEEE Spectrum ranking, Python once again takes first place. JavaScript showed the biggest drop, falling from third to sixth place. Meanwhile, in the separate "Jobs" category, which only considers employer demand, Python also took the lead, displacing SQL. Although the latter remains one of the key skills on developer resumes.
The ranking methodology is based on a combination of open data: Google search queries, discussions on Stack Overflow, GitHub activity, and mentions in scientific publications. However, over the past two years, the volume of such signals has sharply declined. More and more programmers are turning to ChatGPT or Claude instead of asking public questions on forums, while assistants like Cursor write routine code themselves. As a result, the number of new questions on Stack Overflow in 2025 was only 22% of the previous year's level.
Because of this, measuring language popularity is becoming increasingly difficult. But more importantly, the very necessity of choosing a language is gradually losing its significance. If previously the syntax, features, and rules of a language were critical, now these tasks are increasingly being handled by AI. Programmers are starting to argue less about where to put a semicolon or which indentation is more correct—and are focusing more on architecture and algorithms.
AI is capable of generating code in almost any language, provided there is training data. This calls into question the future emergence of new languages: if previously their promotion was aided by books, articles, and demo projects, this is insufficient for AI. It requires large volumes of code for training, and therefore less common languages find themselves at a distinct disadvantage.
In the long term, this could "freeze" the popularity of existing languages. Launching new projects will be more difficult, and the choice of language will increasingly become a technical detail, much like the specifics of particular processors once were.
Some researchers are already questioning whether high-level languages are needed at all. If AI can directly convert a programmer's request into intermediate code for the compiler, then traditional languages could turn into an unnecessary layer of abstraction. However, this would lead to the creation of "black boxes" that are impossible to read but can be tested in modules.
The role of the programmer in such a future will also change. Architectural decisions, the choice of algorithms, system integration, and working with new hardware will remain important. That is, fundamental knowledge will be valued more highly than proficiency in a specific language.
Whether there will be a "main programming language" in 2026 is a big question. One thing is clear: AI has already become the most significant factor for change in development since the advent of compilers in the 1950s. And even if some of the current hype around AI turns out to be a bubble, the practice of using LLMs for writing code is here to stay.
IEEE promises to search for new metrics and approaches in the coming year to understand what "language popularity" even means in the age of AI.
