Torras.ai

Intelligence is a slippery concept, especially when it comes to comparing it between humans and machines. While we might excel in certain areas, AI has been improving its capacity to understand and solve complex problems by leaps and bounds. But can it truly match the human mind?

One way to tackle this question is by challenging both humans and AI to grapple with a problem that requires more than just raw computational power. It's about the ability to comprehend, analyze, and reason through intricacies—a hallmark of human intelligence.

I vividly recall a moment during my college days when my professor threw an intriguing riddle at us. It was one of those puzzles that seemed simple at first glance but unfurled layers of complexity as we delved deeper. It took me and my classmates some time, but eventually, the solution clicked into place.

Now, let's examine this riddle:

The King and the Two Wizards

A small kingdom had a king who regularly received counsel from two very smart wizards. Though the two wizards were truly wise, they didn't show much interest in the daily affairs of the country.

One day, the king grew fed up with the wizards and locked them in a tower in the middle of his kingdom. He informed them that, as a test of their wisdom, they would be kept in solitary confinement in the tower—one in a room facing east and the other in a room facing west.

Each would see half the kingdom (though not necessarily half the number of cities in the kingdom) and would be able to count the cities in the half that they could see. Between them, they would be able to see all the cities in the kingdom, but neither would see any of the cities that the other could see. Additionally, they could not communicate with each other.

Also, neither of them could partially see a city; any city they could see, they could see in its entirety.

Every evening, a jailer would visit them in turn, and ask them about the total number of cities in the kingdom. If either of them could tell the jailer the total number of cities in the kingdom, they would both be released that same evening.

However, if either of them provided the wrong answer, they would both be executed. The poor wizards had no idea about the number of cities in the kindgom.

The last piece of information the king gave them before they were taken to their separate rooms was that there were either 12 or 13 cities in his kingdom.

The first night, the wizards didn't say anything when the jailer asked. The next night, they stayed silent again. Days went by, and they just kept quiet.

However, on the eighth evening after they were locked in the tower, one of wizards finally spoke. His answer was correct, and as a result, they were both set free.

So I have these questions for you:

• How many cities were there in the small kingdom?
• How many cities did each see?
• How did the wizards know how many cities there were in the small kingdom?

Curious about how well our current AI technology fares in comparison to human intelligence, I decided to put ChatGPT (or GPT-3.5) to the test. Surprisingly, while it managed to guess the total number of cities correctly about 50% of the time (as expected due to its binary nature), it consistently stumbled when it came to answering the other two questions embedded within the riddle.

Enter GPT-4o, the latest iteration of AI prowess by OpenAI. With heightened anticipation, I posed the same riddle to this new model, repeating the experiment multiple times. The results were astonishing: GPT-4o failed to provide a correct answer even once.

This brings us to the ultimate question: Can you outsmart GPT-4o? Are your cognitive faculties sharper than the most advanced AI on the market? It's time to put your problem-solving skills to the test and see where you stand on the intelligence spectrum.

And don't hesitate to contact me if you want to know the answer to this riddle!