The Curious Case of AI's Time Blindness
In a surprising revelation that highlights the peculiar limitations of artificial intelligence, leading chatbots like ChatGPT and Claude have demonstrated an inability to perform one of the most basic human tasks: telling the current time. While these AI models can browse the web, write complex code, and analyze images, they stumble when asked "what is time right now."
When questioned by TOI Tech desk, ChatGPT responded with a candid admission: "I understand what you're asking — but I don't have access to a real-time clock, so I can't see the current time on your device or anywhere else." The chatbot instead recommended users check their phones, laptops, or other devices for accurate Indian Standard Time (IST).
Which Chatbots Pass the Time Test?
The time-telling capability varies significantly across different AI platforms. While ChatGPT and Anthropic's Claude failed the test, several other major chatbots demonstrated perfect time awareness. Google's Gemini, Microsoft's Copilot, and Elon Musk's xAI-owned Grok successfully displayed the exact time along with time zone details and additional information.
Claude's response mirrored ChatGPT's limitations, stating: "I don't have access to real-time information, so I can't tell you the current time." The chatbot suggested alternative methods including checking devices, asking voice assistants like Siri or Alexa, or looking at nearby clocks.
The Technical Challenge Behind AI's Time Blindness
According to AI robotics expert Yervant Kulbashian, who extensively studied AI's perception of time in 2024, the problem lies in how language models operate. "A language model works in its own space of language and words. It is only referencing things that have entered this space," Kulbashian explained in a statement to The Verge.
The expert used a compelling analogy: "It's like a castaway on an island in the middle of the ocean, stocked with a massive collection of books but no watch." While OpenAI could technically build a bridge to provide ChatGPT access to system time, this creates significant technical challenges.
Kulbashian detailed the core issue: ChatGPT has limited space in its context window - the information it remembers at any given time. Each time the AI checks a system clock, it consumes valuable space in this window. "If you start adding more things onto your desk, you have to eventually start pushing things off," he noted, comparing constant clock updates to someone interrupting a conversation every minute to announce the time.
Researchers Demonstrate Workarounds and Risks
Pasquale Minervini, a natural language processing researcher at the University of Edinburgh's School of Informatics, demonstrated how ChatGPT could indeed tell time when given proper access. His desktop ChatGPT app immediately showed the correct time in Milan, Italy, where he was located.
"It's able to tell the time if you give it access to a clock. Otherwise, it's something that was just born in that moment, in a way," Minervini explained. The timely information was likely "embedded in the context" of the app when he activated the "Search" function, allowing ChatGPT to use his computer's built-in time tools.
However, Minervini warned about the risks of allowing AI models to search the web or access personal data, including exposure to malicious prompts online. His research also revealed that AI models struggle with many time-related tasks, including reading analogue clock hands correctly and understanding calendars.
OpenAI's Official Stance and Future Improvements
OpenAI spokesperson Taya Christianson addressed the time-telling limitation directly: "The models powering ChatGPT don't have built-in access to the current time, so for up-to-date facts ChatGPT sometimes needs to call search to pull in the latest information."
Christianson confirmed that "We're continuing to improve how consistently it knows when to do so," indicating that future versions might handle real-time information more effectively. This acknowledgment suggests that OpenAI recognizes the limitation and is working toward solutions.
The time-telling challenge represents a fascinating intersection of AI capabilities and limitations, demonstrating that even the most advanced artificial intelligence systems still struggle with tasks humans consider fundamental. As AI continues to evolve, bridging these basic functionality gaps will be crucial for creating truly intelligent assistants that can handle both complex computations and simple, real-world queries with equal proficiency.
