Google pilots AI‑assisted software engineer interviews

Google will let software‑engineering candidates use its Gemini AI model during the “code comprehension” interview round, beginning in the second half of the year as part of a pilot for junior and mid‑level roles in the United States [2].

Pilot design and interview changes

The internal document reviewed by Business Insider describes the new format as “human‑led, AI‑assisted.” Candidates will still be overseen by interviewers, but they may query Gemini while analysing existing code rather than writing solutions from scratch [1]. Interviewers will assess “AI fluency,” including prompt‑engineering, output validation and debugging skills [2]. The pilot will first run in Google’s Cloud business and Platforms & Devices unit, with plans to roll out to other orgs and regions if the test proves effective [1][3].

Why Google is shifting its hiring process

Google says three‑quarters of new code generated inside the company came from AI as of April [2]. Industry peers such as Canva and AI‑coding startup Cognition already allow AI use in technical interviews, arguing that prohibiting tools is akin to banning calculators on a math test [2][3]. By aligning interview tasks with the “GenAI era,” Google aims to evaluate candidates on managing and improving AI‑generated code—a skill set increasingly central to modern software development [1][3].

What to watch

• Rollout milestones – Expansion beyond the initial US teams to other divisions and global locations.
• Candidate performance metrics – How AI‑fluency scores compare to traditional coding‑only assessments.
• Industry response – Whether other large tech firms adopt similar AI‑assisted interview formats.

If the pilot confirms that AI‑augmented interviews better predict on‑the‑job performance, Google could set a new hiring standard for engineering roles, forcing competitors to rethink how they evaluate coding talent in an AI‑driven development landscape.