We're preparing to recruit Windows-based systems administrators and usually conduct interviews remotely because the roles themselves are remote. During our last hiring round, it became increasingly difficult to tell whether candidates were answering from their own knowledge or receiving real-time help from AI tools or another person. We want to keep the process accessible to remote applicants without repeating that experience. What interview techniques or verification steps have worked for making remote technical interviews more reliable?
5 Answers
Focus on whether the candidate can communicate naturally and demonstrate judgment. Ask about previous incidents, have them explain trade-offs, and probe details that are not in the initial question. Someone may use AI for preparation or transcription without that being a problem; the important distinction is whether they personally understand the advice they would give customers and can recognize when an automated answer is wrong.
There isn't a foolproof technical trick. The most reliable approach is to use remote interviews for initial screening, then have shortlisted candidates complete a final interview with identity verification in person. That could be at your office, a local HR partner, a staffing agency, or a booked meeting space. It adds cost and may exclude some legitimate applicants, so limiting it to the final stage is probably the best compromise.
Video, screen sharing, and eye contact are not enough on their own. Candidates can use another device, an earpiece, or software that listens to the conversation and generates suggestions. Locked-down equipment can reduce some risks, but it still cannot eliminate phones or outside assistance. Treat those measures as supporting controls, not proof that the interview is secure.
Sending a controlled device is possible, but it can be expensive and inconvenient, and a phone could still be nearby. A supervised final assessment is generally more dependable.
Make the technical discussion feel like a real troubleshooting conversation instead of a list of predictable questions. Ask the candidate to work through a scenario, interrupt occasionally with new information, and follow up on the choices they make. For example, start with a slow application report, then add details about DNS, authentication, resource usage, or recent changes. Someone who has actually done the work can usually adapt naturally, while scripted or generated answers become much harder to maintain.
Open-ended questions and quick follow-ups are much harder to answer with a prepared prompt than standard question-and-answer interviews. You are evaluating their reasoning and communication, not just whether they reach the expected conclusion.
Use remote interviews to narrow the pool, but include a practical assessment for the finalists. Give them a realistic troubleshooting task and ask them to explain their decisions as they work. You can also repeat or vary part of the task later to confirm that the person understands what they submitted. The goal should be validating practical ability rather than trying to detect a particular AI application.
A local HR or staffing partner can simply host the candidate while your team conducts the interview remotely. It has also helped organizations verify location and prevent someone else from taking the interview.