Leaderboard

Since the shift to online video interviews, the opportunity for candidates to leverage AI tools for assistance has increased. With sophisticated language models now available, cheating in virtual interviews is becoming more common. Here are some insights and practical tips on spotting AI-aided responses and preventing potential misuse during your online interviews: Recognizing Potential AI Use Delayed Responses with Flawless Answers: Candidates using AI assistance often have a noticeable pause before delivering well-crafted answers. Interviewees who genuinely know the answer usually respond naturally without long pauses, often thinking aloud or talking through their reasoning. If you detect a pattern of pauses followed by near-perfect answers, it could indicate AI involvement. Unusual Overviews and Irrelevant Tangents: AI tools frequently start responses with a brief topic overview or unrelated details before answering. In contrast, real candidates tend to answer the question directly. If a candidate begins each response with background information that wasn’t asked for or strays off-topic, it might raise a red flag. Trust your instincts if something feels off. Fixed Eye Movements: People’s eye movements are typically dynamic, glancing at the screen, occasionally down, or to the side when thinking. However, candidates relying on AI tools might stare fixedly at one spot, as though reading from another screen. Notice if the candidate's gaze is less natural and more "fixed" during responses. Tips to Prevent AI Assistance in Interviews Use Open-Ended, Scenario-Based Questions: Avoid questions with straightforward answers. Instead, provide scenarios that require candidates to discuss their reasoning and understanding. This approach encourages authentic responses. Frequent Follow-Up Questions: If you sense anything unusual, ask follow-up questions immediately. AI-assisted responses are less adaptable, and candidates may struggle to answer off-the-cuff follow-ups meaningfully. Ask “Why” and “What Makes You Think…” Questions: Questions that dig deeper into reasoning can help determine if the candidate truly understands the topic or is just repeating rehearsed answers. Display Questions Visually: Try sharing your screen or presenting questions on the screen during the interview. This can throw off candidates using voice-to-text tools, which might require them to refocus unexpectedly. Incorporate Visuals or Diagrams: Use diagrams or visual aids, then ask specific, related questions. Responses derived from AI may miss nuanced details in visual prompts, helping you differentiate genuine understanding. These methods can help you maintain interview integrity and identify authentic responses. Remember, subtle cues often reveal much, so trust your instincts!

Retrieval-Augmented Generation (RAG) and fine-tuning for LLM-Powered Agent (Large Language Models) are both types of AI language generation methods of Natural Language Processing (NLP) with some fundamental difference in response generation. Let's understand: In RAG, the response generation is based on external source of knowledge that helps with real-time information update & the same information is used to augment the model's response. Basically, instead of relying on internal knowledge used for model training, RAG has kind of "Open-Book" approach where any additional or real-time information is looked from external sources. Whereas in Fine-tuning for an LLM-Powered Agent, there is pre-existing internal data source which is fine-tuned through continuous improvement on a specific set of data to specialize it for specific task or domain. Say for example, we are adding specific books in internal data sources for generation of more advanced & niche responses which industry or domain specific. Application for RAG: Dynamic & real-time knowledge requirements (e.g. stock price, latest news, latest research etc.) Model Size Reduction (as no need for huge internal data source required) Cost & Efficiency (requirement of low or no internal data eliminates requirement of high energy consuming servers for storing internal database) Application for Fine-Tuning with LLM Powered Agent: Nuances & Specialized Tasks (e.g. application with special domain language, knowledge and jargons like medical, legal etc.) Consistent Tone and Style (nuanced in line with ask in the prompt) Hybrid Use Case (RAG & Fine-Tuning with LLM Powered-Agent) In cases where we need the response that is specialized in language and also dynamic with respect to real-time or latest information update, we need both internal and external sources for response generation. In this scenario, we need to explore the Hybrid model. Examples: Let's say we have created Financial Advisor Chatbot which uses finance related language and tax laws using (Fine-Tuning with LLM) as well as gives advice considering latest market dynamics (stocks, bonds, investment avenues etc. - RAG). In this case, it's best to use both internal and external sources to generate ideal response. To summarize, hybrid model is ideal scenario for specialized and real-time response generation.