Google is introducing a new assessment method that allows the use of artificial intelligence assistant tools in software engineer hiring interviews.

Business Insider reported on May 7 that Google is preparing a pilot program for some teams in the United States that would allow the use of AI assistant tools in interviews for junior to mid-career developers.

The focus is the "code understanding" interview. Google plans to allow the use of approved AI assistant tools from the second half of this year. Applicants will be tasked with reading an existing codebase, finding errors and improving performance. The tool used in the pilot phase is Google's own AI model, Gemini.

The evaluation method will also change. Interviewers will not look only at simple coding results. An internal document states that interviewers assess "AI usage proficiency," including prompt writing, output validation and debugging skills. Google described the overhaul as "to better fit the modern engineering environment."

Brian Ong (브라이언 옹), a vice president in charge of hiring at Google, said, "Google continues to evolve our interview process to hire the best talent." He said the pilot is an attempt to reflect in interviews how the real working methods of software organisations have changed in the AI era.

The overhaul is expected to expand across the interview process. The "Googleyness and leadership" interview, previously centred on behavioural questions, will add a technical design discussion based on applicants' past projects. For some junior applicants, an interview that solves an open-ended engineering task will be used instead of one round of the existing technical interview.

Google plans to test the new interview method first from this month across several departments, including its cloud and platforms and devices organisations. If results are confirmed, it plans to expand to other regions and organisations. For now, the initial rollout is limited to some teams in the United States.

The backdrop is change in development work itself. Google said in April that three quarters of newly written code internally is being generated by AI. OpenAI President Greg Brockman (그렉 브록먼) also recently mentioned that the share of code written by AI has increased from 20 percent to 80 percent. The fact that Anthropic and OpenAI released new models late in 2025 that significantly improved coding-agent performance is part of the same trend.

Similar moves are already emerging in the tech industry. Canva and AI coding startup Cognition also allow AI use in technical interviews. Cognition's head of people and operations, Emily Cohen (에밀리 코언), said barring AI use in interviews is "similar to making a child take a math test without a calculator." She also said that when building something similar to real work, AI tools can be used and should be used.

Google's pilot shows that standards for what counts as skill in developer hiring are changing. It reflects a direction in which hiring processes begin to treat not only the ability to write code directly but also the process of using AI, validating results and correcting errors as job competency.