Should AI Tools Be Allowed In Engineering Interviews?

AI coding assistants have evolved extremely rapidly, fundamentally changing how engineers work. As tools like Claude Code, Cursor, GitHub Copilot and ChatGPT enable us to generate production-grade code in seconds, hiring teams face an important question: Should candidates be allowed to use AI tools during engineering interviews?

Companies like Meta and Canva are proponents of redefining their technical interviews to incorporate the use of AI tools, while companies like Amazon strictly prohibit the use of AI tools, emphasizing the need to ensure a level playing field. Anthropic, which used to strictly prohibit the use of AI tools, now allows AI in parts of its process with clear guidance.

So the answer to the question isn’t a simple “yes” or “no.” While effective use of AI tools is important in modern software development, allowing the candidate to freely use them comes with the risk of an inability to assess the candidate’s fundamental technical skills. Getting this balance right is critical when hiring engineers in this modern age of AI.

The Importance Of Fundamental Technical Skills

AI isn’t perfect, with AI tools prone to hallucinations and subtle bugs, and they often fail on edge cases, thereby requiring significant human validation.

A 2025 study comparing human vs. AI code found that AI-generated code is more repetitive and contains more high-risk defects, while another experiment showed that after multiple rounds of AI self-improvement, critical vulnerabilities increased by 37.6%. Further, according to a recent randomized field study involving experienced open-source software developers, the use of AI tools resulted in a 19% increase in the time it took the developers to complete their tasks because they had to spend extra time validating and fixing its output.

It is therefore essential for engineers to have solid core technical skills so that they can actually fix such issues whenever they arise.

The Challenges Of Allowing AI Tools In Interviews

Allowing the use of AI tools in technical interviews makes many standard interview questions trivial. Modern AI tools can answer technical conceptual problems or solve classic algorithmic problems with elegant code in a matter of seconds. If the candidate is using AI tools to address such interview questions, a strong prompt engineer can easily appear to be a strong developer, even if they are not.

In order to meaningfully assess a candidate’s real-world development ability while allowing the use of AI tools, the assignment needs to be a large, open-ended project that cannot be completed by mere prompting. However, preparing and reviewing such large-scale projects requires significant time and effort from the hiring team.

A Balanced Solution: Separate Assessments For Core Skills And AI Fluency

A balanced solution would be a hybrid approach with separate assessments for core technical skills and AI fluency. Pair-programming sessions, provoking deep dives (i.e., asking candidates to explain the rationale behind their choices, how they would handle edge cases, etc.) and using proctoring tools that ensure the candidate is not utilizing AI tools can effectively assess the candidate’s core technical skills.

Upon successful assessment of these skills, the hiring team can provide a separate assignment or interview session that allows the candidate to use AI tools. During this process, the hiring team needs to assess how the candidate prompts, validates the AI output, handles mistakes that the AI makes and refactors the code. An important red flag to look for would be the candidate blindly copying and pasting the output from the AI tool.

Assessing Engineering Talent In The Age Of AI

AI tools are changing the landscape of software engineering, thereby causing the engineer hiring process to evolve along with it. While AI fluency is valuable in modern software development, it should complement, not replace, the assessment of core engineering skills.

Originally published at https://www.forbes.com.