The New Digital 'Proxy Interview'
The scene is a familiar one for millions of Indian engineering graduates: a high-pressure, remote technical interview for a coveted tech job. The difference in 2026 is the invisible assistant in the room. Candidates are leveraging generative AI tools
like ChatGPT and Gemini as real-time copilots. With a problem statement pasted into a chat window on a hidden phone or second screen, a complete code solution can appear in seconds. This phenomenon goes beyond simple cheating; it's a sophisticated process that some have dubbed the 'new-age proxy interview'. Candidates use pre-rehearsed prompts and specialised tools that provide instant, structured answers not just for coding challenges, but also for complex system design and behavioural questions. The scale of this trend is significant; one tech hiring platform noted that it flags around 30-35% of sessions in Indian university recruiting for suspicious, AI-assisted behaviour.
Decoding the 'System Prompt'
The headline's term, 'system prompt', has a specific meaning in the world of AI. It refers to the background instructions that developers give an AI model to define its personality, rules, and constraints before a user even types a word. While most graduates aren't rewriting the source code of ChatGPT, they are using a similar principle: creating their own structured, pre-defined prompts designed to reliably extract high-quality, interview-passing answers from AI. This can involve telling the AI to act as a 'senior software engineer', to explain concepts simply, or to format code in a specific way. It's less about a single-shot question and more about developing a system to guide the AI towards the perfect answer under the constraints of an interview.
A Response to Brutal Competition
The impulse to use AI as a co-pilot isn't born in a vacuum. For many, it's a pragmatic response to a hyper-competitive job market. One cloud architect noted that the pressure is so intense that many see it as a necessary shortcut rather than an act of laziness. With thousands of applicants vying for a handful of positions, online technical assessments can feel like an impersonal and sometimes unfair barrier. Some candidates argue that if the job itself will involve using AI coding assistants—which 82% of professional developers now experiment with—then using them during an interview is just a reflection of the modern work environment. This has sparked a debate: where is the line between using a tool to enhance your skills and having the tool replace them?
The Recruiter's Counter-Offensive
Companies are acutely aware of this trend and are engaged in a technological arms race to preserve the integrity of their hiring process. Major assessment platforms like HackerRank and HackerEarth have massively increased the use of AI-powered proctoring, which analyzes webcam feeds and screen activity to flag suspicious behaviour like frequent tab-switching or code being pasted instantly. Many have moved assessments from the browser to secure desktop applications that block unauthorised software. The most effective countermeasure, however, has proven to be a low-tech one: the follow-up question. Recruiters report that a candidate who has relied on AI can rarely explain the nuances of the code they supposedly just wrote, failing within minutes when asked to elaborate on their logic.
Is AI Proficiency the New Skill?
As the cat-and-mouse game evolves, a more progressive idea is taking hold. Some forward-thinking companies are shifting their perspective entirely. Instead of trying to ban AI, they are starting to embrace it as part of the evaluation. Companies like Meta and Shopify are piloting interviews where candidates are expected to use AI assistants. The assessment then shifts from pure coding ability to what they call 'AI judgment'. Can the candidate craft an effective prompt? Can they spot and debug errors in AI-generated code? Do they know when to ignore the AI and trust their own knowledge? In this new model, prompt engineering isn't cheating; it's a core competency, and the ability to effectively collaborate with an AI is the very skill being tested.
