AI Malfunctions: ChatGPT Struggles with Basic Time-Telling

ChatGPT, the AI developed by OpenAI, has been criticized for its inability to accurately tell time. Users have reported experiences where the chatbot provides either correct answers, incorrect ones delivered with confidence, or declines to answer altogether.

This inconsistency raises concerns about the reliability of AI systems, especially given ChatGPT's advanced capabilities in web browsing, coding, and image analysis. AI robotics expert Yervant Kulbashian highlighted the fundamental limitations of large language models, stating that these AIs operate like a castaway stranded on an island filled with books but lacking a watch.

They predict answers based on historical training data without real-time updates, which is crucial for tasks like telling time. OpenAI has the ability to give ChatGPT access to system clocks through features like Search, yet this functionality comes with significant trade-offs.

Each clock check consumes a portion of the model's context window, the limited space it can use to process information at any given moment. Additionally, Pasquale Minervini, a natural language processing researcher at the University of Edinburgh, pointed out that leading AI models struggle not only with telling time but also with interpreting analog clocks and managing calendar-related tasks.

This situation underscores a broader issue within the AI industry regarding the trust and functionality users can expect from these systems. The implications of these limitations could impact how users interact with AI in the future as they grapple with the balance between sophisticated capabilities and fundamental accuracy in basic tasks.

The report raises essential questions about the future of AI technologies and the potential risks involved in relying on them for everyday functions.