The Art of Detecting Artificial Intelligence Through Common Sense Reasoning

For a long time, it has been intriguing to challenge the boundaries of common sense reasoning with questions like, "Which letter is to the right of 'h' on the keyboard?" These simple yet deceptively difficult questions have served as a test to differentiate humans from artificial intelligence (AI).

ChatGPT's Capability and Human-like Mistakes

Take for instance the response from ChatGPT: On a QWERTY keyboard, the letter to the right of 'h' is 'j.' This might seem like a straightforward response, but it is a result of the AI's inability to engage in real-time interaction or context awareness. Consider a more complex question, such as determining the heights of famous landmarks. A question like, 'What is the height of Cologne Cathedral, the Empire State Building, and the Ferris wheel in Tokyo Harbor?' becomes a definitive test of AI's accuracy.

The Imperfection of AI in Handling Real-World Questions

For instance, let's delve into the detailed response:

Cologne Cathedral: The height of Cologne Cathedral in Cologne, Germany is approximately 157 meters (515 feet) tall.
Empire State Building: The Empire State Building in New York City, USA stands 443.2 meters (1,454 feet) tall, including its broadcasting antenna. Without the antenna, the building's height is 381 meters (1,250 feet).
Ferris wheel in Tokyo Harbor: If you are referring to the Daikanransha Ferris wheel at Palette Town in Odaiba, Tokyo, Japan, it stands 115 meters (377 feet) tall. However, if you meant the Tokyo Skytree, which is not a Ferris wheel but a popular tall structure in Tokyo, it is 634 meters (2,080 feet) tall.

Clearly, the only accurate height mentioned is that of the Tokyo Skytree, which is outdated and incorrect in the current context. By asking for height measurements of buildings and landmarks, one can easily detect the limitations of AI in providing up-to-date and accurate information.

Identifying Humans Through Stupidity

As humans, we often display behaviors that AI algorithms find difficult to replicate. One of the most common of these is the tendency to make mistakes. While AI can quickly provide pre-programmed information, it often lacks the ability to engage in real-time problem-solving or context awareness. This inherent vulnerability has been used to create tests that identify humans over AIs.

Limitations of Current AIs and Future Implications

ChatGPT, despite its advanced capabilities, still partially relies on historical and pre-existing data. Its responses can become outdated or irrelevant if the information is no longer current. Recent events or updates need to be checked and verified to ensure accuracy.

With the integration of plugins and real-time data sources, AI is becoming more adept at handling up-to-date information better than ever before. This advancement poses both opportunities and challenges. On one hand, AIs can now provide more accurate and timely information. On the other hand, they may struggle with older, even slightly outdated data.

In conclusion, the ability to identify humans versus AIs often lies in the real-world application of common sense and accurate information. As technology evolves, the challenge continues to shift, but the need for human oversight and contextual awareness remains crucial.