\

Outsmarting Machines: The Riddle AI Can’t Solve, but Humans Can

  • Jordi Torras
  • Blog

In May 2024, I penned an article titled Are You Smarter Than GPT-4o? It featured a riddle I’ve long cherished: The King and the Two Wizards. The origins of this riddle trace back to my days as a student, when a professor shared it during a class. The puzzle challenged us, sparking a lively debate that culminated in a solution—though it took considerable effort and collaboration to crack it.

I’ve kept the answer under wraps intentionally. It’s a personal secret, and I’m determined to shield it from AI’s ever-watchful algorithms. However, if you’re curious, I’d be delighted to share it with you—just reach out to me.

Back in May, I decided to test GPT-4’s reasoning abilities with this riddle. The results were lackluster. Later, when Claude 3.5 launched, I tried again, only to be met with disappointment. OpenAI’s much-anticipated o1 model? It failed spectacularly. Each attempt revealed that, despite their astonishing capabilities, these AI systems faltered when faced with this nuanced problem.


Teaching and Testing the Riddle

Teaching at EADA Business School in Barcelona, I’ve revisited this riddle as part of my Software Engineering and AI lectures. I recount it exactly as I heard it decades ago, challenging a new generation of students to grapple with its intricacies. The responses are always enlightening. About two-thirds of the class typically zone out, perhaps distracted by their phones. Most of the rest struggle to comprehend the problem, but there’s always one who stands out.

This year, that student was exceptional. Before I could even finish explaining the riddle, she confidently declared, "This is easy." To everyone’s astonishment, she proceeded to the blackboard, presented her solution, and defended it eloquently. In that moment, I witnessed a flash of human intelligence that left me awestruck. It was a reminder of why I love teaching and why I believe in the boundless potential of the human mind.

What Sets Us Apart

Instances like this reinforce my conviction: human intelligence, while imperfect and inconsistent, remains uniquely brilliant. The challenges of the future will demand more of this brilliance than ever before. Yet, as AI advances, it’s likely to leave some people behind—those who rely too heavily on machines without nurturing their own capacities.

I’m eagerly awaiting the release of OpenAI’s o3 model. Will it succeed where its predecessors failed? If it does, I’ll be impressed. If it doesn’t, I’ll remain equally fascinated, pondering how long it will take for AI to achieve this level of reasoning. Either outcome will deepen my appreciation for the evolving interplay between human and artificial intelligence.

Looking Ahead

2025 promises to be a landmark year for AI. Stay tuned—there’s much more to come. And remember: while machines may dazzle us, it’s the light of human intelligence that truly illuminates the path forward.

Make AI work for you

Empower your vision with our expertise. Me and my team specialize in turning concepts into reality, delivering tailored solutions that redefine what's possible. Let's unlock the full potential of AI. Effectively.

Contact us