Google Revolutionizes Development: 75% of Code Generated by AI

Google recently revealed that three-quarters of its new code is now generated by artificial intelligence. This proportion has seen steady growth, with the company encouraging its employees to adopt AI tools to enhance their productivity.

Google's CEO, Sundar Pichai, emphasized that the use of AI agents has enabled code migration to be completed six times faster than before. This advancement demonstrates the increased efficiency that AI can bring to development processes.

AI at the Heart of Development at Google

Currently, 75% of the code produced at Google is generated by AI, although each line is subsequently reviewed by human engineers. In October 2024, this figure was 25%, and it had already reached 50% the previous year. This rapid progression illustrates Google's commitment to integrating AI into its workflows.

Sundar Pichai mentioned in a blog post the transition to "truly agentic workflows," allowing engineers to work more autonomously thanks to AI.

Collaboration and Internal Tensions

A complex code migration has recently been successfully completed through collaboration between AI agents and engineers, resulting in a sixfold increase in speed compared to the previous year. Google's engineers are using the Gemini models to generate code, and some have specific AI-related goals that will influence their performance evaluations.

However, the use of Anthropic's Claude Code by some employees at Google DeepMind has sparked internal tensions, according to reports from Business Insider.

Trends in the Tech Industry

Google is not alone in this endeavor. Microsoft, for example, has already integrated AI into its development processes, with 20 to 30% of the code for certain projects generated by AI, according to its CEO Satya Nadella. Microsoft's CTO, Kevin Scott, predicts that 95% of the code could be generated by AI in the next five years.

Meta is also pursuing this path, aiming for 55% of code changes to be assisted by agents by the end of 2025, and 65% of engineers are expected to use AI for more than 75% of their code by 2026.

Snap recently announced that at least 65% of its new code is generated by AI, highlighting a growing trend in the industry to rely on artificial intelligence for software development.