Yandex developer relations director: AI in code is an exoskeleton, not autopilot

Sergey Berezhnoy, director of developer relations at Yandex and expert in Yandex's online master's program "Frontend, Backend Development and AI Solutions" in partnership with Yandex Praktikum, published a column on Habr about why creativity and the ability to think have become the key skills of a developer in the era of generated code.

Why AI is an exoskeleton, not an autopilot

The central thesis of the column: AI in development should be perceived as an exoskeleton that amplifies human capabilities, not as an autopilot capable of replacing his decisions. An exoskeleton requires that a person continues to choose the direction himself and evaluate the result — the model only accelerates and facilitates execution, but does not relieve the developer of responsibility for architectural and product decisions.

Why a developer using AI is already a team lead

Berezhnoy formulates another thesis: if a developer uses AI in development, he is already effectively performing the role of a team lead. Working with a code-generating model requires the same skills as managing a team — you need to set a task, check the result, find errors in someone else's (albeit generated) work and decide what to keep and what to rewrite. The author emphasizes that this is precisely why creativity and critical thinking become more important, not less important, as the model takes on routine code writing.

• Author — Sergey Berezhnoy, director of developer relations at Yandex
• Expert in the online master's program "Frontend, Backend Development and AI Solutions" from ITMO and Yandex Praktikum
• Key thesis: AI in development is an exoskeleton, not an autopilot
• Using AI in code effectively turns a developer into a team lead for the model

What this changes in the profession of developer

If developer value was previously measured primarily by the speed and accuracy of code writing, then as code-generating models spread, other qualities come to the forefront — the ability to formulate a task so that the model understands it correctly, the ability to quickly evaluate the quality of a generated solution, and the willingness to take responsibility for the final result, even if the code itself was not written by human hands.

What this means

As code generation becomes a routine task for AI models, the developer's role shifts from executor to task setter and quality controller — effectively to the role of team lead, even if formally there are no people in his subordination.