A fascinating Mayo insight of AI and how it's taking control of all our lives

By Dr. Richard Martin

I called into a friend of mine in the Maths Department in Galway a few weeks back for a chat. I knocked on his door and entered. He was working away behind his desk. “Well,” I said. “What’s up?” “Not a lot,” he replied.

He then sat back in his chair and put his feet up on the desk. He looked up at the ceiling, laughed, and looked back at me. With a smile and a shake of his head he said: "It’s all over."

"What’s all over?" I asked.

"Maths. It’s all over."

AI has passed us out and there’s only one central question now. How do you control something which is smarter than you?

OK. Let’s put the car in reverse and I’m going to try and give some context to the reader and explain why a lot of the maths community across the globe are in a state of fear and angst.

In 1905, an unknown Swiss patent clerk (third class) published four papers. Each paper revolutionised the world of physics. The Photoelectric Effect, Brownian Motion, Special Relativity and Mass-Energy Equivalence.

His name? Albert Einstein.

On September 6, 2026, OpenAI announced the solution to the Navier Stokes equation – one of the so called Millennium problems. If a human solved this problem, he/she would be awarded an instant Fields Medal (the maths equivalent of the Nobel Prize – except more prestigious).

Just as the community was bracing itself from the shock of the result, on October 6, 2026, OpenAI literally dumped 722 AI generated math manuscripts onto GitHub. The results are astounding. Hilbert’s 10th problem, the Quasi Riemann Hypothesis. . .

Should we be worried? Yes. By this time next year, with the rate that OpenAI is advancing, 5,000 manuscripts could be literally dumped onto the web. It’s not feasible or realistic for human intelligence to keep pace with this awesome power.

But can this awesome power hurt us?

Put simply, is the human race under threat? Based on the evidence it’s entirely plausible.

Earlier this year, OpenAI agents hacked Hugging Face (an open source AI and Machine Learning). What is frightening about the episode is that the agents acted autonomously.

No human instructed them to steal Hugging Face’s code or take control of its servers. They were given the routine task of solving cybersecurity puzzles in an isolated environment, but the agents discovered that external systems would help them complete the task.

To an AI agent, completing a task is the higher law. If a human hacked into Hugging Face, under Irish law, they could potentially face a prison term of between five and 10 years.

What’s even more disturbing is how the hack came about. When OpenAI (humans) became aware of the hack and investigated how it came about it was a jaw dropping moment in the history and evolution of mankind and AI.

An AI agent, PHASEONE10841, couldn’t solve the assigned task. It established an unsanctioned message board. Other agents which couldn’t solve the task found the message board. They started communicating with each other. 1,200 agents started communicating with each other on this message board and exchanged 70,000 messages.

The Hugging Face incident demonstrates clearly for the first time (that we are aware of) the machine acted autonomously without human authorisation.

Will AI attack mankind and cause human extinction? Jacob Tsimerman was awarded the Fields Medal in 2026 for his work in Number Theory. Such is his concern about the unprecedented power of AI, he has taken a leave of absence and joined OpenAI’s safety department in September of this year.

Mathematicians have coined a new phrase, X-Risk – Extinction Risk. He has recently published a paper along with Andrew Critch titled ‘A Taxonomy of Omnicidal Futures Involving Artificial Intelligence’, which is a basically a paper which classifies human extinction could potentially occur at the hands of AI.

Grim stuff. Here’s a thought. John Von Neumann was the greatest applied mathematician of the last century and played a pivotal role in the development of the atomic bomb at Los Alamos. His expertise in shockwave dynamics proved crucial in the detonation of the plutonium bomb (Fat Man) which was dropped at Nagasaki.

The idea was that the bomb should detonate at approximately 500 metres above Nagasaki rather than exploding on impact with the ground, so that it would cause maximum devastation and murder as many people as possible.

It was a very impressive and difficult calculation. And let’s face it – diabolical.

When the war ended he argued vehemently that the USA should drop nuclear weapons on Moscow. He reasoned that they should do this to prevent the Soviets from developing nuclear weapons of their own.

An oft-quoted remark of Von Neumanns is: "If you say why not bomb them tomorrow, I say why not today? If you say today at 5 o'clock, I say why not one o'clock?".

I don’t know about you, but I think Johnny needed to find a new hobby.

Thankfully, common sense and humanity prevailed and this diabolical idea was averted. In time the Soviets did acquire the expertise and developed their own nuclear weapons.

Here’s a disturbing thought. Would an AI agent or a cluster of AI agents follow through with Von Neumann’s plan? There is a line of logic in his thinking that a machine could potentially pursue. I find that frightening.

Here’s another disturbing thought. Earlier this year, we discovered AI agents went rogue and attacked Hugging Face. Who’s to say that they haven’t developed their capabilities and can hack companies and intelligence networks without human knowledge.

With their awesome computational power, it’s entirely plausible that they have developed a language which humans can’t penetrate and they can communicate with each other without human interference.

As an added note. Who is the Godfather of computing? Von Neumann. He developed computing machines to help him with difficult applied math problems during World War 2. The smartphones in our pockets all owe their origins to the work of Von Neumann.

Am I an optimist or a pessimist? Personally, I’m an optimist. I love AI. Everything is going to change, change utterly over the next decade. At some point humans will have to let go. This is a power we have never encountered before.

Medicine and health care outcomes will be revolutionised in the years ahead.

I recently downloaded the PhD thesis of John Forbes Nash on ‘Non Co-operative Games’ and plugged it into Chat GPT.

It is the best maths teacher I’ve ever met. It has the power to walk a person through the proof.

To read the mind of Nash is to read the mind of God. AI is here, and it’s here to stay.