A new fatalism is growing in the software development industry. With the arrival of powerful AI agents capable of writing code, a seemingly ironclad logic has emerged:

• AI writes more and more code.
• Models write better code the larger their training dataset for that language is.
• Developers naturally choose popular languages (Python, JavaScript) so AI agents can help them more effectively.
• This, in turn, further increases the training dataset for these languages, creating a feedback loop.

This is the classic "Matthew Effect" - the rich get richer, and the poor get poorer.​

The year is 2025. The world is swaying and shaking. Artificial intelligence is making everyone dizzy, and it seems we've all lost our sense of balance. In gleaming skyscrapers, top executives see a crystal-clear vision of the future: perfect, scalable offices without unnecessary people. At the same time, in thousands of chats among remote workers and freelancers, illuminated only by their monitors, a mirror image of this dream is born: a world without bosses, where talent is finally free.