Google is expanding the use of artificial intelligence in its software engineering recruitment process, allowing candidates to use AI coding assistants during interviews while testing systems that use its Gemini models to transcribe conversations, take notes, and help interviewers prepare candidate assessments.
The changes, detailed in internal documents reviewed by Business Insider, reflect a broader shift in how technology companies evaluate engineering talent as AI tools become increasingly embedded in software development. Rather than assessing candidates solely on their ability to write code independently, Google is placing greater emphasis on their ability to work effectively with AI systems and integrate the technology into products.
The company has been introducing the changes in phases throughout 2026. Although internal documents indicate that the revised interview formats have expanded beyond their initial testing groups, Google says the changes remain part of a pilot programme.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
“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, Google’s vice president of recruiting, told Business Insider.
The shift is significant because recruitment interviews help determine which technical skills employers consider essential. By permitting AI assistance during coding assessments and explicitly evaluating candidates’ AI fluency, Google is signaling that competence increasingly involves knowing how to direct, assess, and improve machine-generated work, rather than relying exclusively on unaided coding ability.
Google’s updated hiring framework includes a new “code comprehension” interview designed to assess candidates’ understanding of AI-assisted software development, including their ability to use prompt engineering and work with Google’s Gemini models.
Candidates can use Gemini as an assistant during the interview, allowing recruiters to evaluate how they approach coding problems with AI support. This differs from conventional technical assessments, which often focus on an applicant’s ability to solve problems by writing code independently under time constraints.
The company initially tested AI-related interview changes within a small number of organizational units. Business Insider reported in May that the pilot included assessments of AI fluency and prompt engineering. The internal documents now indicate that the new formats have been introduced more broadly, although Google has not described the programme as a completed company-wide rollout.
A Google spokesperson said AI fluency is a requirement for all software engineering positions. AI product integration is also required for most roles outside entry-level positions, according to the documents.
That suggests that Google is evaluating two related but separate capabilities. AI fluency concerns whether candidates can use AI tools effectively, while AI product integration involves applying AI within products or broader technical systems.
The latter requirement is deemed necessary for engineers expected to build services that incorporate AI models, connect them to existing software, or translate model capabilities into useful features. In such roles, technical competence increasingly extends beyond writing code to understanding how AI systems behave within a larger product.
However, allowing AI during interviews also creates a different assessment challenge. Recruiters must distinguish between candidates who understand the code and can evaluate AI-generated suggestions and those who depend on the tools without being able to identify errors, security weaknesses, or inefficient solutions.
However, the changes are consistent with Google’s own adoption of AI-assisted software development. In April, the company said that three-quarters of new code created at Google was generated by AI and reviewed by human employees. That figure illustrates the extent to which AI has become part of the company’s software production process. It also helps explain why Google is adjusting recruitment standards to reflect the way its engineering teams operate.
As AI coding assistants take on more of the work involved in generating code, engineers may spend a greater proportion of their time defining problems, reviewing suggestions, testing outputs, and correcting errors. These activities require technical knowledge, even when the initial code is produced by a machine.
Google has also been encouraging employees to use its Gemini models to improve existing products and develop newer versions of the technology. The hiring changes could help the company recruit engineers who can work in this AI-assisted environment.
But the broader implications extend beyond Google. Technology companies that integrate AI into their development processes face pressure to update the skills they assess during recruitment and the expectations they place on existing employees. AI proficiency is expected to become a baseline workplace competency, much as familiarity with established development tools became standard in earlier stages of the industry’s evolution.
Yet the shift raises questions about how employers should assess junior engineers. Traditional coding interviews can test foundational programming knowledge and problem-solving ability, but allowing AI changes the balance between unaided technical execution and the ability to supervise a tool.
If employers place too much weight on AI-assisted performance, they could struggle to distinguish strong underlying engineering skills from an applicant’s ability to prompt a model. Conversely, assessments that ignore AI may fail to reflect the actual working environment candidates will enter.
A more effective approach would evaluate both: if candidates understand core programming concepts and if they can use AI responsibly to improve their work.
Gemini Moves into The Interview Process Itself
Google is also testing Gemini for administrative tasks associated with recruitment. One internal document describes a pilot involving AI-assisted transcription and note-taking during some job interviews conducted through Google Meet.
The company is also working to enable interviewers to use Gemini to draft feedback tailored to the specific role being filled.
These applications are expected to reduce the time recruiters and technical interviewers spend documenting conversations and preparing assessments. They may also help standardize feedback by prompting interviewers to connect their observations to the requirements of a particular position.
However, AI-generated notes and feedback introduce their own risks. Transcription systems can misinterpret statements, while generated assessments may omit context or place undue emphasis on particular answers. Interviewers will need to verify the material and retain responsibility for their evaluations to avoid allowing automated summaries to distort hiring decisions.
Google’s internal guidance also draws a boundary around the use of AI in recruitment. Employees participating in the pilots were warned not to use generative AI tools to create their own standalone solutions or agents for hiring and interviewing applications.
“Do not use GenAI or LLMs to create your own standalone solutions or agents for hiring or interviewing use cases,” the document states, referring to generative AI and large language models.
The restriction indicates that Google’s willingness to introduce AI into recruitment does not amount to unrestricted permission for employees to build independent hiring tools. The company is testing specific applications within a controlled process while limiting unauthorized implementations.
The reason is that recruitment involves sensitive candidate information and decisions that can affect people’s careers. Any wider use of AI in hiring must account for confidentiality, accuracy, consistency, and the risk that automated systems could influence decisions in ways interviewers do not adequately understand.
Google’s evolving process ultimately points to a change in the definition of engineering competence. As AI becomes a routine part of software development, employers are increasingly interested in how well candidates can combine technical knowledge with AI assistance. The challenge is to ensure that the technology strengthens the assessment of engineering ability rather than obscuring the difference between genuine expertise and effective tool use.