Let’s be real – AI is everywhere now. Tools like ChatGPT, Gemini, and other AI chatbots are transforming the way candidates prepare for interviews. In IT recruitment, this means you might not always be hearing the candidate’s authentic experience; sometimes you’re getting a polished AI version.
As one of the best IT recruitment agencies in Poland, we’ve noticed a growing trend: candidates using AI to craft “perfect” answers. And while AI can be helpful, it can also make it tricky to figure out who really knows their stuff. The good news? With a few advanced techniques, you can spot the difference and still run effective, human-centric interviews.
This guide is designed for experienced recruiters who want to stay ahead of the curve, spot AI-assisted answers, and ensure they’re truly evaluating a candidate’s skills and experience.
Understanding the AI Factor in Interviews
Candidates use AI in two main ways:
- Preparation: Using AI to prep answers to common interview questions.
- Real-time assistance: Accessing AI during video interviews or coding sessions (less common but not impossible).
Signs of AI-assisted responses include:
- Answers that are overly structured, polished, or “textbook.”
- Lack of personal anecdotes or concrete examples.
- Responses that are too formal or generic.
- Technical explanations that don’t align with the candidate’s claimed experience level.
AI doesn’t think – it predicts. So, while it can generate correct answers, it often lacks the nuance, depth, and personalization that comes from real-life experience.
Advanced Questioning Techniques
The goal is simple: ask questions AI can’t answer convincingly. You want depth, context, and personal touch.
- Behavioral & Story-Based Questions
Ask about specific past experiences. Example:
- “Tell me about a time you debugged a system failure under pressure. Walk me through the process and how you decided on the solution.”
- “What’s a technical decision you made that didn’t go as planned? How did you handle it?”
AI can generate stories, but candidates struggle with details, emotional nuance, and reflection when pressed.
- Follow-Up & Probing
Don’t stop at the first answer. Dig deeper with “why/how” chains:
- “Why did you choose that approach?”
- “What alternative methods did you consider?”
- “How did this impact your team or project?”
This exposes shallow, AI-generated answers, because AI often provides only the surface-level rationale.
- Time-Boxed Reasoning
Ask candidates to think on the spot. For example:
- Give them a short problem to solve live.
- Ask for pros and cons of a decision in 90 seconds.
- Have them draw a quick diagram or write pseudocode.
AI-generated answers are slow to reproduce live, and candidates relying on it may hesitate or revert to generic responses.
- Personalized & Context-Specific Questions
Ask questions that require unique insight or personal context:
- “If you were joining our backend team, how would you improve the current API workflow?”
- “Based on our recent product launch, what would you optimize and why?”
AI can’t account for your specific environment without a lot of setup.
Detecting AI-Generated Responses
Here’s a practical cheat sheet for spotting AI answers:
- Language cues: overly formal, structured, repetitive phrasing.
- Generic answers: lacks examples, personal insight, or nuance.
- Depth mismatch: technical explanations too perfect or misaligned with claimed experience.
- Behavioral inconsistencies: tone or detail varies unnaturally between questions.
Combine these checks with live coding, pair programming, or problem-solving sessions. A candidate can talk theory, but demonstrating it in real-time is much harder to fake.
Deepfake Detection in Video Interviews
AI isn’t just in text – it’s creeping into video too. Deepfake technology can simulate candidates or alter video responses. While rare, it’s worth knowing what to look for.
Red flags in video interviews:
- Microexpressions that don’t match speech.
- Eye movements or blinking that look unnatural.
- Lip sync inconsistencies or delayed reactions.
- Unnatural body movements or lighting discrepancies.
Tools to help detect deepfakes:
- Microsoft Video Authenticator – detects subtle manipulation.
- Deepware Scanner – identifies deepfake videos.
- Sensity AI – analyzes suspicious facial patterns.
- Amber Video – verifies video authenticity.
Practical tips for interviews:
- Ask candidates to interact with props or perform quick, on-camera tasks.
- Request random actions like writing a note, coding on-screen, or showing hands.
- Multi-angle verification can catch inconsistencies.
Sample Interview Flow
Here’s a structure combining everything above:
- Warm-Up – Light behavioral questions to ease into conversation.
- Deep Dive Behavioral – Ask for detailed stories, follow-up with why/how chains.
- Technical Assessment – Live coding, problem-solving, or pair programming.
- Context-Specific Questions – Ask about your company/project context.
- Consistency Checks – Compare responses across sections for contradictions.
- Closing Personalization – Wrap up with a question on personal goals or opinions.
This multi-layered approach makes it difficult for AI to cover all angles convincingly.
Closing Thoughts
AI is here to stay, and it’s changing the recruitment landscape. But human insight, intuition, and real-time interaction remain irreplaceable. As one of the best IT recruitment agencies in Poland, we know that adapting your interview techniques is key to spotting genuine talent and ensuring that your hires truly fit.
Remember: AI can assist, but it cannot replicate authentic experience, creativity, and the human touch. By asking the right questions, probing deeper, and leveraging both technical and behavioral assessments, you’ll stay ahead of the curve.
Stay sharp, keep innovating, and treat AI as a tool – not a threat.
Having a problem with candidates overusing AI in your recruitment process? Let’s have a chat
Autor – Adam Łyko, CEO Sowelo Consulting
