In November 2024, a stunning claim began to circulate across social media. Hieu Pham, an engineer at Elon Musk’s AI company xAI, posted something that grabbed attention worldwide.
Grok 3, their latest and most advanced AI model, had supposedly proven the Riemann Hypothesis. This is one of the greatest unsolved problems in mathematics.
The post even stated that training Grok 3 was paused to verify the proof. The AI had become so powerful it might pose a danger to humanity.
It sounded like the plot of a sci-fi movie. But could it be real?
Let’s investigate what really happened. And whether Grok 3 is actually capable of cracking a 160-year-old mystery.
What Is the Riemann Hypothesis, and Why Is It So Important?
First, let’s break down what the Riemann Hypothesis is, in simple terms.
Proposed in 1859 by mathematician Bernhard Riemann, the hypothesis deals with the distribution of prime numbers. It’s tied to something called the Riemann zeta function and suggests that all the “non-trivial zeros” of this function lie on a specific line where the real part equals 1/2.
If that sounds complex, here’s the impact. Proving this would unlock deeper knowledge about prime numbers. These numbers are crucial in everything from modern cryptography to quantum physics.
It’s so significant that it was named one of the seven Millennium Prize Problems. Each is worth $1 million to anyone who can solve them.
Despite decades of effort, the Riemann Hypothesis remains unsolved. That’s why the idea of an AI proving it sounded unbelievable and revolutionary.
How Good Is Grok 3 at Math?
To understand whether Grok 3 could even attempt something like the Riemann Hypothesis, we have to look at what it is. And what it can do.
Grok 3 is xAI’s most advanced chatbot. It’s designed to compete with ChatGPT, Claude, and Google’s Gemini.
But unlike those models, Grok 3 emphasizes math, science, and reasoning over general conversation.
It was trained on 200,000 GPUs. It has real-time internet access and includes features like Big Brain Mode and Think Mode for deeper reasoning.
In 2024, Grok 3 scored 52 points on the AIME (American Invitational Mathematics Examination). That is no small feat.
Its smaller version also competes closely with models like Claude 3.5 in reasoning tasks.
In short, Grok 3 is a serious contender in math-heavy domains.
Did Grok 3 Actually Prove the Riemann Hypothesis?
Let’s rewind to mid-November 2024.
The internet lit up when a post from Hieu Pham, an engineer at xAI, claimed that Grok 3 had done something no human had accomplished in over 160 years. According to the now-viral message, the AI had successfully proven the Riemann Hypothesis, one of the toughest problems in mathematics.
The post didn’t stop there. Pham added that training for Grok 3 was immediately paused, as the model had reached such an advanced level of reasoning that it posed a potential danger to humanity.
Needless to say, people were stunned.
Grok-3 just proved Riemann's hypothesis. We decided to pause its training to check its proof, and if the proof is correct, training won't be resumed, as the AI is deemed so smart that it becomes a danger to humanity.
— Hieu Pham (@hyhieu226) November 17, 2024
Twitter (now X) exploded. Memes started flooding in. Mathematicians, AI experts, and regular users all chimed in. Some were skeptical. Others were excited. A few simply wanted to believe it was true.
Could this be the moment where AI crossed the final frontier? Not just mimicking intelligence, but actually creating groundbreaking mathematical knowledge?
But then came the twist.
Just hours later, Pham clarified that it was all a joke. A tongue-in-cheek response to growing excitement around Grok 3’s capabilities. There was no proof. There was no breakthrough. And certainly no pause in training due to AI becoming self-aware or dangerous.
Alright, Saturday Night Life is over.
— Hieu Pham (@hyhieu226) November 17, 2024
As for why a proof to Riemann's hypothesis is dangerous, I highly recommend @matthaig1's wonderful novel: The Humans.https://t.co/NBF80AJEJ1 https://t.co/qmb328ytW9
Instead, it had started as a casual comment that spiraled out of control. Like many viral stories in the age of instant reaction, what began as sarcasm got taken way too seriously.
Other xAI team members leaned into the humor. Greg Yang, a co-founder of xAI, sarcastically joked that Grok 3 had begun “attacking the security guards in the office.” Another researcher, Heinrich Kuttler, claimed they had to reset the AI by replacing all its bad weights with “NaN,” short for Not a Number, a common placeholder in computing when something goes terribly wrong.
Even the AI community online joined in, with parody posts, humorous threads, and exaggerated theories. Some likened the incident to a modern-day Turing Test. Not for AI intelligence, but for how easily humans could be convinced (or confused) by AI hype.
But beneath all the humor and memes was a deeper takeaway.
People took the claim seriously not just because it was shocking, but because it felt possible.
Why This Joke Hit So Hard
Think about it: we live in a world where AI models write essays, pass medical exams, generate art, and even compose music. They diagnose illnesses, help design rockets, and translate ancient texts.
So when someone says, “This new AI just solved an unsolvable math problem,” it doesn’t sound completely ridiculous anymore.
That’s what made this moment so powerful. For a split second, people around the world thought: “Maybe it happened. Maybe we’re really there.”
Even though the Riemann Hypothesis remains unsolved, the story sparked serious conversation. Experts began asking real questions: How close are we? Can AI actually help? Is it possible in the next few years?
And more importantly: What happens when it really does happen?
Could AI Actually Prove It Someday?
This is where things get exciting.
While Grok 3 hasn’t cracked the Riemann Hypothesis, the idea of AI solving unsolved math problems isn’t science fiction anymore.
Google DeepMind’s AlphaProof recently solved several problems from the 2024 International Mathematical Olympiad. That includes the hardest one.
Its reasoning resembled human intuition. It wasn’t just calculation.
And Terence Tao, a Fields Medal winner, believes that by 2026, AI will be trusted as a co-author in mathematical research.
That means AI wouldn’t just assist with math. It would collaborate with mathematicians to tackle abstract problems.
Still, there are hurdles.
AI lacks true conceptual understanding. It doesn’t “know” math in the human sense.
It processes patterns, logic, and symbols without intuition. Solving something like the Riemann Hypothesis requires both brute force and creative insight.
Right now, AI can help verify proofs. It can suggest strategies and analyze complex logic. But it can’t yet deliver a fully original solution to something as deep as the Riemann Hypothesis.
The Power and Limitations of Grok 3
What Grok 3 offers is speed, access, and structure.
It retrieves real-time information. It reasons through complex problems and even interacts with users in a more human-like way.
It’s the first AI with direct access to X and live data. This makes it very powerful for fact-checking, coding, and math-heavy tasks.
It’s not perfect. It still hallucinates facts sometimes.
It’s available only to X Premium+ users. Its ability to do original research is still developing.
But the fact that it even tries to tackle problems like the Riemann Hypothesis puts it in a class above traditional chatbots.
As Andrej Karpathy noted, Grok 3 approaches these problems with boldness. Unlike other models that simply say, “This problem is unsolved,” Grok 3 actually tries.
AI Solving Unsolved Math Problems: What Comes Next?
So, did Grok 3 prove the Riemann Hypothesis? No. It was a joke. A clever one that sparked real conversations.
But it revealed something deeper. We’re entering an era where AI isn’t just answering questions. It’s trying to solve them.
Whether it’s Grok 3 or another future model, the dream of AI solving the crown jewels of mathematics is becoming more realistic by the day.
And when that moment comes, when AI truly proves something like the Riemann Hypothesis, it may not just be a breakthrough in math.
It may be the day we rethink what intelligence really means.
