Google just announced that over a quarter of their new code is written by AI. During their Q3 2024 earnings call, CEO Sundar Pichai shared this update, which raises some important questions about the future of software development.
Here’s what we know: Engineers still review and approve all AI-generated code before it goes live. This helps catch potential errors, copyright issues, and other problems that could cause system failures.
The benefits are clear – AI handles repetitive coding tasks while engineers focus on more complex problems. But there’s a catch. Engineers now spend time fixing AI mistakes and learning how to write better prompts, which creates its own time sink.
Pichai didn’t specify which AI tools generate the code. He did mention their Gemini chatbot and Notebook Language Model for note-taking. They’ve also expanded Google Search’s AI Overviews to over 100 countries, though users have reported incorrect information and annoying ads in these features.
This matches what I’ve seen in my consulting work. AI is adept at many tasks, but it still sometimes makes mistakes. I advocate for a semi-automated approach, where AI assists human developers but doesn’t entirely replace their oversight and expertise.
