Google isn't just asking if you can code anymore — they're asking if you can make an AI code for you.
The Summary
- Google is rolling out AI-integrated interviews for software engineers, testing candidates on prompt engineering and "AI fluency" as baseline skills
- New "code comprehension" format lets candidates use Gemini during interviews, shifting evaluation from pure coding ability to AI-assisted problem solving
- 75% of new code at Google is now AI-generated and human-reviewed, making the traditional coding interview increasingly irrelevant to actual job requirements
The Signal
Google's interview revamp tells you everything about where software engineering is headed. The company started piloting AI-integrated interviews in May with a handful of teams. Now it's company-wide. The new format includes a "code comprehension" interview where candidates are evaluated on their ability to work with AI coding assistants, specifically prompt engineering skills.
Candidates can use Gemini during the interview itself. Think about what that means. Google is explicitly saying: we don't care if you can write perfect code from memory under pressure. We care if you can effectively direct an AI to write it.
"We're piloting updates to our interview process for software engineers, including evaluating for AI fluency, to be more reflective of how our teams are operating."
The stat that matters: Google reported in April that 75% of new code at the company is AI-generated and reviewed by humans. Not 10%. Not 30%. Three-quarters. The job isn't writing code anymore. It's reviewing code, refining prompts, and making architectural decisions that AI can't make yet.
Interviewers can now use Gemini to generate feedback for candidates. That's worth noting not because it's technically impressive but because it shows how far into the workflow AI has penetrated. Even the meta-process of evaluating engineers is being automated.
Here's the deeper signal: Google is ahead on this because they have no choice. They built the tools. They saw internal productivity metrics shift before anyone else. Now they're hiring for a different job than the one that existed 18 months ago. The "software engineer" role is morphing into something closer to "AI whisperer with code literacy."
The Implication
If you're interviewing anywhere in tech over the next year, expect this. Google moves, others follow. The LeetCode grind isn't going away completely, but it's becoming table stakes, not the main event. The new question is: can you work with AI to solve problems faster than someone who can't?
For hiring managers, this is your playbook update. Test for AI fluency now or hire people who'll be obsolete by the time they finish onboarding. For engineers, the writing is on the wall. Learn prompt engineering. Get comfortable treating AI as a junior dev on your team. The next interview might not test if you can build a balanced binary tree. It might test if you can get Gemini to do it correctly on the first try.