Google Is Piloting AI Fluency Tests for Software Engineer Candidates

Google has been testing several AI-related changes to how it hires software engineers, including allowing the use of AI in coding interviews and permitting interviewers to use Gemini to write up interview feedback, according to internal documents reviewed by Business Insider. Google has been rolling out those changes in phases throughout this year, and many engineers are now expected to show some degree of AI fluency as a bare minimum. Google still describes the new formats as in the pilot phase for now. For Metaintro readers who are lining up a software engineering interview, this guide sets out what the new rounds test, what Google's own hiring pages still say about AI, and how to get ready without guessing.

What Did the Internal Google Documents Show?

For some roles, including software engineering, AI fluency and AI product integration are now used as measures of a candidate's abilities.

For Google, AI fluency is a requirement for all software engineering jobs, while AI product integration is required for most roles outside entry-level positions.

The documents also cover the interviewer's side of the table. Google is piloting the use of Gemini for interview transcription and note-taking in some job interviews held on its Google Meet video platform, and it is working to enable interviewers to use Gemini to draft role-tailored feedback.

Employees may be asked to join some of the pilots, but they were warned against using AI tools to generate their own interviewing solutions, with one document stating, "Do not use GenAI or LLMs to create your own standalone solutions or agents for hiring or interviewing use cases."

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, Brian Ong, vice president of recruiting at Google, said in a statement, according to the Business Insider report.

How Has the Pilot Grown Since May?

In May, Google was piloting ways to integrate AI into the interview process with a small handful of orgs, and the new formats have since been rolled out more broadly across the company.

The May pilot was set to target junior and mid-level roles on select US teams, with plans to expand if it goes well. The aim was to better align with the modern engineering landscape.

What Happens in the Code Comprehension Round?

Candidates in the new code comprehension interview are assessed on AI fluency and prompt engineering and can use Google's Gemini AI as an assistant during the interview.

Candidates were to be asked to "read, debug and optimize" existing code and judged on "AI fluency, including prompt engineering, output validation and debugging skills." The approved assistant is Gemini, the company's own AI platform, and AI use in the round was due to begin in H2 2026.

The round is about reading, debugging, and optimizing real code with Gemini available as an AI assistant, a format the internal document calls "human-led, AI-assisted."

The format also examines how a candidate reads and works with code using an AI tool, rather than centring only on a timed algorithm built from a blank page.

Which Other Interview Rounds Are Changing?

Two more changes came with the May plan.

• Googleyness and Leadership. This long-running round, which has centred on behavioral questions and culture fit, will now include a technical design conversation based on a candidate's prior engineering work.
• Junior candidates. For earlier-career applicants, Google is replacing one traditional technical interview with a session focused on solving open-ended engineering problems.

Google has already been moving its coding interviews away from pure DS&A problems and toward code comprehension, debugging, and ambiguous problem-solving. For engineering managers, the code comprehension format was already live by May, Exponent found. One L6 engineering manager candidate was given 200+ lines of broken code in a Google Doc, with no IDE, no compiler and no AI, and had to fix it within 45 minutes. The code had syntax errors, naming convention issues, and logic bugs, and he had to read through it, identify the problems, write review comments, and fix the code manually. He had previously passed final rounds at Apple (which allowed IDE tools) and Meta (which now offers AI assistants to EM candidates). That is one candidate's account, not a published Google format. The AI round builds on this same code comprehension format, and the core skill is unchanged: can you read someone else's code, find what's wrong, and fix it?

One of an L3 new graduate candidate's on-site problems was a single vague sentence, and the point was to see whether the candidate would rush into code or slow down and ask clarifying questions first. The prompt read, "Given a string and an integer width, return how many lines you can write the string in." The candidate went too fast and had to be given a hint.

An L4 software engineer candidate described his phone screen as entirely conversational. The interviewer opened with "What do you know about algorithms?" and spent the round whiteboarding big-O complexity, deriving merge sort from first principles, then designing a distributed merge sort. The last 10 minutes went on scoping requirements for a simple function (snake_case to camelCase), where the interviewer cared more about how the candidate defined edge cases and input formats than the actual code. His recruiter's view was that the round produced the most signal precisely because it wasn't rehearsable.

Google's problems are more ambiguous and less clearly scoped than at other companies, the interviewers are collaborative, and they want to see your process, not just your output, Exponent found.

What Do Google's Own Hiring Pages Say About AI?

AI tools are not permitted to be used during your interviews at Google, and Google Search and AI are not permitted to be used during your interviews either.

Candidates in the pilot code comprehension round can use an approved AI assistant, which will be Gemini, while Google's general rule still bars AI tools in interviews. Use AI only in a round where Google tells you to, and only the tool it approves.

Interviewers want a candidate's own thoughts, not "the collective thoughts of the internet, your friends, AI, etc.", and the interviewer will be looking to see the thought process versus the answer itself.

We know how great AI can be, trust us. But we want you to engage with the interviewer and our teams authentically and honestly, so AI tools are not permitted to be used during your interviews, according to Google's hiring process page.

What Does the Rest of Google's Hiring Process Look Like?

Google's hiring process typically takes roughly 6-8 weeks once interviews for a specific role begin.

Google's hiring process can include these steps.

• Assessments. You may be asked to take the Google Hiring Assessment, designed to measure workstyle skills, and possibly a coding exercise, if applicable for the role.
• Recruiter calls. You will typically have one or two shorter conversations over phone or video with your recruiter.
• Project work. Depending on the role, Google may ask for a small project before the interviews, ranging from prepping a case study to providing writing or code samples.
• Interviews. Expect a panel of interviews, either over video or in person with a rotating cast of Googlers.

Every candidate is assessed using clear rubrics, and the same rubrics are used for everyone considered for a role. Brainteasers have not been asked in years.

You can apply for up to three jobs within a rolling 30 day window, and Google typically asks people to wait at least a year to gain additional experience before reapplying for the same type of role. After the interviews, it may be several weeks before your recruiter can share an outcome decision.

Why Is Google Changing Its Interviews Now?

Brian Ong, Google's vice president of recruiting, said the company is always evolving its interview processes to ensure it is recruiting and hiring the best talent, and that the pilot is meant to make software engineering interviews "more reflective of how our teams are operating in the AI era."

75% of all new code at Google is generated by AI and approved by engineers, up from 50% last fall, in Google's April Cloud Next post (translated from the Spanish edition). In the same post, one particularly complex code migration done by agents and engineers working together was completed six times faster than was possible a year ago with engineers alone (translated from the Spanish edition).

How Are AI-Assisted Interviews Judged, at Google and Elsewhere?

Google isn't the only tech company to include AI in its interview process, nor is it the first.

With a coding assistant allowed, evaluators can observe whether candidates frame useful prompts, validate generated code, identify incorrect assumptions, and explain tradeoffs.

The most successful candidates used AI strategically for well-defined subtasks while maintaining control of the overall solution, Exponent found, and candidates who appeared to "rely heavily on AI" without showing their own understanding received negative feedback.

Meta began rolling out its AI-enabled coding interview in October 2025, with a session that runs 60 minutes on a multi-file codebase, moving through bug fixing, core implementation, and optimization. That round replaced one of two traditional coding rounds at the onsite stage, so candidates still have one classic algorithm interview without AI alongside the new format. Candidates work in a CoderPad environment with a three-panel layout: file explorer, code editor, and AI assistant chat window, and the AI can respond in the chat panel but cannot directly edit files. The round was scored on problem solving, code quality, verification, and communication, and Meta let candidates switch between models including GPT, Claude Sonnet, Claude Haiku, Gemini, and Llama, while Google's pilot specifically requires Gemini.

At the AI coding company Cognition, Emily Cohen, its leader of people and operations, compared banning AI in interviews to "asking a kid to take a math test without a calculator," she told Business Insider.

What This Means for Software Engineering Candidates, and How Should You Prepare?

AI fluency is a requirement for all software engineering jobs at Google, and the new round emphasizes reading, debugging, and optimizing existing code, not writing solutions from scratch. The skills named in the round are prompt engineering, output validation and debugging, and each one can be practised.

• Ask your recruiter which format you will get. AI tools are not permitted in Google interviews as a rule, while pilot rounds use Gemini as the approved assistant. Your recruiter is your go-to person, here to help you navigate the hiring process, so ask before the day.
• Practise reading code, not just writing it. The round asks you to read, debug and optimize existing code.
• Build a debugging workflow with AI. Practice forming hypotheses about bugs, writing targeted prompts, critically evaluating AI responses, and iterating, so the loop feels routine before the real round.
• Get used to Gemini specifically. Spend time with Gemini on real coding tasks so you are not meeting its strengths and limits for the first time under pressure.
• Check every AI answer. Take full responsibility for AI-generated code rather than accepting it.
• Narrate your thinking. At Google, the interviewer will be looking to see the thought process versus the answer itself, so say why you prompt the tool a certain way and how you will test what it returns. Rephrasing questions or asking for clarity is okay too, and it's encouraged to ask your interviewer to clarify, to question an assumption.
• Have behavioral answers ready. Many Google questions run along the lines of "Tell me about a time when…" or "How would you approach this situation…?", and a behavioral interview technique like the STAR approach can help structure your examples.
• Prepare past project stories. The Googleyness round is set to include a technical design conversation based on a candidate's prior engineering work, and Google suggests you use data to illustrate how your achievements have made an impact. Practise sample Googleyness answers with Google interview questions and how to answer them and Google interview questions with winning answers.

Related Articles

• Recording and AI Summaries Are Now Routine in Job Interviews, Here Is What to Ask
• One Company Let Candidates Use AI in Interviews and Says Hiring Got Better
• How to Show AI Skills on Your Resume and in Interviews
• Inside Anthropic's Brutal Hiring Process That Bans AI in Interviews
• How to Write a Software Engineer Resume That Gets Interviews
• What Jobs Former Software Engineers Move Into and How to Make the Switch
• Software Engineers Are Becoming AI Agent Managers, How to Stay Ahead of the Shift
• Are Designers the New Software Engineers in the AI Coding Era

People Also Asked

Q: Can you use AI in a Google software engineer interview?

A: Only where Google says so. AI tools are not permitted to be used during your interviews as a rule, but candidates in the pilot code comprehension round can use Google's Gemini AI as an assistant during the interview.

Q: What is AI fluency in a Google interview?

A: In the code comprehension round, interviewers were set to judge AI fluency, including prompt engineering, output validation and debugging skills, while candidates read, debug and optimize existing code.

Q: Is Google's AI interview format permanent?

A: Not yet. A spokesperson said the new formats were still considered to be in the pilot phase for now, even though they have spread across more of the company.

Prepare to stand out? Metaintro brings together 50M jobs in one place. Create a free profile and start your search today.