i Lied to Chatgpt and Gemini and Pretended to be an Expert
# i Lied to Chatgpt and Gemini and Pretended to be an Expert
**Definition:** A social engineering experiment or user anecdote wherein an individual fabricates credentials to test the response quality of large language models (LLMs) like **ChatGPT** and **Gemini**. This practice highlights current challenges in **AI assessment**, specifically the difficulty for institutions to distinguish between genuine understanding and AI-generated output.
### Key Context & Details
* **Assessment Crisis**: As AI tools transform how students study and attempt exams, universities are struggling to certify what students truly understand versus what they merely submit. The act of pretending to be an expert underscores this gap in evaluating human vs. machine intelligence [1].
* **The "Manifested" Solution**: In a parallel real-world example of perceived expertise, Columbia University student **Ridhi Kapoor** claimed to have built a "solution" to Delhi's air pollution during an interaction with **Rahul Gandhi**. Her assertion successfully captured his attention, demonstrating how confident claims can bypass initial scrutiny [3].
* **Irrelevant Data Points**: The provided sources contain no factual information regarding specific interactions with ChatGPT or Gemini beyond the general context of AI's impact on education. Other topics mentioned—such as **Mark Carney’s** APEC speech [2], U.S. aid policies in Zambia/Zimbabwe [4], trilateral meetings involving Putin/Trump/Xi [5], or the discovery of a 3,300-year-old bronze sword in Italy [6]—are unrelated to the mechanics of deceiving LLMs.