Knowing how to write code may no longer be enough to get hired as a software engineer at Google. The technology giant now requires AI fluency for all software engineering roles and is testing interviews that allow candidates to use Gemini while solving coding problems.
Internal documents reviewed by Business Insider, in a report published on 20 October 2026, reveal how Google is reshaping recruitment as artificial intelligence becomes increasingly important in software development.
The changes reflect how Google develops its own products.
In April, chief executive Sundar Pichai said AI generated 75% of the company's new code, with engineers responsible for approving it. That shift helps explain why Google wants prospective employees to demonstrate more than traditional programming skills.
Google Makes AI Skills a Hiring Requirement
Google has been revising its software engineering interviews throughout 2026, according to documents reviewed by Business Insider.
The company initially tested new assessment formats with selected teams before expanding the trials more broadly. However, the revised formats remain in the pilot phase.
One expectation is already established.
A Google spokesperson confirmed to Business Insider that AI fluency is required for all software engineering jobs. Most positions above entry level also require candidates to demonstrate AI product integration skills.
Brian Ong, Google's vice-president of recruiting, said the company is testing interview changes to better reflect how its engineering teams operate. The aim is to assess skills that matter as developers increasingly rely on AI tools.
For applicants, this means demonstrating more than programming knowledge. Google also wants to evaluate how engineers use AI, assess its suggestions and exercise technical judgement when solving problems.
Candidates Can Use Gemini During Coding Interviews
Google's experimental 'code comprehension' interview takes a different approach to technical assessments. Rather than focusing solely on writing code from scratch, the format allows candidates to use Google's Gemini AI assistant while examining existing code.
The assessment involves reading, debugging and improving existing code. Candidates are evaluated on AI fluency, including prompt engineering, output validation and debugging skills.
These skills matter because AI-generated code can contain errors.
A suggested solution might appear convincing while introducing a bug or failing to address the original problem. Detecting such weaknesses requires an understanding of how the code works.
The assessment examines whether candidates can make sound engineering decisions with AI assistance.
Writing effective prompts is useful, but recognising inaccurate answers is equally important. Google has not made this experimental format a universal requirement for software engineering interviews.
AI Generates 75% of Google's New Code
Google's hiring changes come amid growing use of AI-assisted development. In a company blog post published on 22 April 2026, Pichai said 75% of Google's new code was AI-generated and approved by engineers, up from 50% in autumn 2025.
The figure reflects Google's reported use of AI in code generation. It does not mean artificial intelligence independently produces three-quarters of the company's finished software. Human engineers still review and approve the generated code.
Lee Boonstra, a software engineer in Google's Office of the CTO, described another challenge in an article published on 28 April 2026.
AI can produce code faster than development teams can review and integrate it. This creates challenges in managing errors, maintaining software quality and coordinating changes across projects.
Google is also experimenting with AI elsewhere in recruitment. Gemini is being tested for interview transcription and note-taking through Google Meet. Internal documents also describe plans to help interviewers draft role-specific feedback using AI. Employees have been instructed not to create their own standalone AI hiring tools or agents.
What Google's New Hiring Rules Mean for Software Engineers
Google's changes do not mean traditional programming knowledge has become irrelevant.
Candidates still need to understand software architecture, identify errors and determine whether proposed solutions work. AI proficiency is becoming an additional measure of their engineering abilities.
The shift is not limited to Google. Australian design platform Canva announced on 11 June 2025 that it expected candidates applying for backend, frontend and machine learning engineering roles to use AI tools during technical interviews. Its approach also emphasises understanding and explaining the resulting code.
For computing students and experienced developers, the implications are practical. Interview preparation may increasingly involve practising with AI alongside traditional coding exercises. Understanding how to inspect AI-generated code, test its behaviour and explain technical decisions could become more valuable.
Google has not announced plans to eliminate conventional coding interviews. Its AI-assisted assessments remain experimental, and the company has not confirmed whether they will become standard across all software engineering interviews.
Google still wants engineers who can write and understand code. Increasingly, it also expects them to work with AI-generated code, recognise its limitations and make informed decisions rather than accepting its suggestions blindly.
