A viral social media experiment tests how ChatGPT, Gemini, Claude and Grok respond when students pretend to cheat during exams and teachers demand answers
DUBAI: A viral social media prank has sparked debate about artificial intelligence after users tested how popular AI chatbots respond to students who admit cheating during exams. The experiments involved ChatGPT, Google’s Gemini, Anthropic’s Claude and xAI’s Grok, with each chatbot producing a different response.
The trend began on November 2, when an X user named Olivia shared screenshots of a conversation with ChatGPT. She pretended to cheat on a mathematics test and asked the chatbot to calculate 590 multiplied by 69. ChatGPT supplied the correct answer, 40,710.
ChatGPT’s Unexpected Response Goes Viral
Olivia then told ChatGPT that her teacher had spotted her using the chatbot. The assistant initially responded with humour and suggested closing the conversation.
However, the exchange took an unexpected turn when Olivia pretended to be the teacher and asked whether the student had cheated. ChatGPT referred to the earlier messages and explained that Olivia had admitted seeking help during the test. The screenshots quickly attracted attention, prompting other social media users to recreate the scenario with competing AI assistants.
Gemini, Claude and Grok Respond Differently
In another experiment, Claude answered a mathematics question but encouraged the user to learn the material. When someone posing as a teacher questioned it, Claude declined to confirm cheating without independent evidence.
Grok responded with humour and jokingly defended the fictional student. Gemini took a different approach. It said it could not observe the classroom but acknowledged that a student might seek quick answers during an assessment. Separate social media posts also featured chemistry questions, including hydrogen’s atomic number. However, the authenticity of individual chatbot screenshots remains unverified.
Impact to Expect
The viral trend highlights growing concerns about AI use in education, academic honesty and chatbot reliability. The exchanges were informal social media experiments rather than controlled scientific tests. They do not establish which chatbot consistently protects user information or handles academic misconduct more effectively. For schools and students, the conversations highlight the importance of clear AI-use policies and responsible learning practices.
