In Google’s recent Q3 earnings call, CEO Sundar Pichai revealed a surprising statistic: over a quarter of the company’s new code is now generated by artificial intelligence. Reporting on Google’s increase in revenue and share value, Pichai shared that AI contributes significantly to coding at Google, though it’s not without its challenges. He noted that 25 percent of Google’s code is created by AI before engineers review and refine it. This, he suggested, helps Google’s developers “do more and move faster.”
However, relying on AI for coding isn’t flawless. Programming assistants powered by AI often insert bugs into code, unintentionally infringe on copyrights, or even trigger system outages. While Pichai didn’t explicitly address these pitfalls, his acknowledgment of the review process implies that human intervention remains essential to catch and correct errors from the AI-generated code.
Some critics argue that this process may not necessarily boost efficiency. After all, Google engineers still spend valuable time revising AI-generated code, which leads some to wonder if direct human coding would be faster and more reliable. Pichai didn’t specify which Google AI products were contributing to code generation, but he mentioned Google’s recently launched Gemini chatbot and Notebook Language Model (LM), an AI note-taking tool, during the call.
Beyond code generation, Google’s AI-powered Search enhancements are driving user engagement worldwide. Pichai emphasized that new features in Google Search’s AI Overviews have been rolled out to over a hundred countries, claiming that these additions increase user satisfaction and search volume. Yet, the rollout of AI-powered results has faced criticism, as users report inaccuracies and increased ad clutter.
Despite mixed reactions to its AI implementations, Google’s investors appear optimistic. As AI transforms coding and product offerings, Google continues to experiment with the balance between AI and human oversight in its ever-growing technology landscape.