I checked whether GPT-5 is superintelligent. It is.
Back on May 14, I published this article titled “Are you smarter than GPT-4o?”. In it, I presented a riddle — a tale I call The King and the Two Wizards.
It’s not your average riddle. To solve it, you need serious reasoning: you must imagine multiple scenarios, think recursively, use inductive logic, and even consider how others would reason in complex, nested situations. In short, it’s a test not of knowledge, but of deep cognitive ability.
Naturally, I wanted to see how AI would fare.
I tried with GPT-4, OpenAI o1, OpenAI o3 and o4 mini, Claude Opus, Mistral, DeepSeek, and others. None of them could solve it. They tried—sometimes generating pages of tangled logic—but ultimately failed. Surprisingly, Grok gave the correct answer once… but only once, and never again.
And then came GPT-5.
Today is August 2025. GPT-5 just came out, and I couldn’t resist: I asked it the same riddle.
The result? Perfection. On the first try. And it took just 49 seconds:.
Let me put that in perspective. When I was a university student studying Artificial Intelligence in the 1980s, this riddle was the kind of challenge that led to days of discussion among my classmates and professors. Back then, one of the most exciting “new” technologies was something called neural networks. I remember coding backpropagation in the C programming language on an Intel 8086 CPU — staying up late just to get the math right and see it run.
Little did I know that, decades later, the same fundamental concept — artificial neural networks, scaled and refined into the deep learning architectures behind GPT — would not only tackle real-world tasks, but also crack the same riddles we used to debate as students.
So, here I am. I’ve officially run out of riddles, mental challenges, logic puzzles — you name it. GPT-5 can solve them all.
Does this mean we've reached AGI — Artificial General Intelligence?
Not quite. GPT-5 still doesn’t play chess like a grandmaster. It can’t drive a car through city traffic like Waymo. But maybe that’s not what “general” intelligence is really about. Maybe AGI is about being good enough — good enough to handle 95% of the intellectual work businesses require: writing, analysis, strategy, planning, coding, communication, decision-making.
And if that’s the case, maybe AGI is already here!.
With GPT-5, we’re entering a new era of AI — one driven not by hype, but by practicality and effectiveness. It’s no longer about proving that AI can be smart. It’s about putting that intelligence to work — across industries, teams, and tools.
And I’m here to help.
Whether you're curious about what's possible or ready to integrate AI into your business processes — reach out!. It’s time to move from testing intelligence to putting it into action.
