What the Infinity Machine taught me about leading in the age of AI in business

The uncomfortable truth about artificial intelligence is not that it might destroy us. It's that the people building it know the risks better than anyone and are building it anyway, accelerating the role of AI in business.

I just finished The Infinity Machine by Sebastian Mallaby, a book I'd strongly recommend to any leader navigating the current AI moment and shaping their AI strategy.

It's a portrait of Demis Hassabis, the chess prodigy, neuroscientist, game designer, and Nobel Prize winner who co-founded DeepMind and has spent his career trying to reverse-engineer human intelligence into a machine. It is brilliant, unsettling, and more relevant to the questions facing Australian business leaders than most management books published this decade.

Here is what stayed with me.

Intelligence Is Not Where You Think It Lives

Hassabis's most provocative idea is that the capabilities we consider uniquely human; intuition, insight, pattern recognition, the kind of understanding buried deep in the hippocampus, are not mystical.

They are learnable. They are reproducible in silicon, given enough data and the right architecture, which is reshaping machine learning in business.

This is not a theoretical position. AlphaFold, DeepMind's protein-mapping model, solved a problem that had defeated biological science for fifty years: predicting the 3D structure of proteins from their amino acid sequences.

It predicted the structure of 200 million proteins. It won Hassabis a Nobel Prize in Chemistry in 2024. The machine did not just compute, it perceived. It found patterns invisible to human researchers who had spent decades looking at the same data.

For leaders managing complex operational environments, where pattern recognition in data, inventory, customer behaviour, and supply chains is the actual competitive advantage across retail operations, this matters. A lot.

"The same capability that decoded proteins can decode your operations. The question is whether you're building the conditions to use it within your AI operations."

The AI Governance Problem No One Wants to Own

The part of The Infinity Machine that should give business leaders pause is not the capability of the tech. It is the AI governance.

Mallaby chronicles how Hassabis evolved from AI utopian to wearied realist, watching the release of ChatGPT demonstrate that once the technology is in the world, no amount of calls for caution will stop the risks from being taken.

The three concerns, that stood out for me from the book:

1. Denial that large language models are powerful enough to cause serious harm,

2. A misplaced belief that fixing the technology itself is sufficient,

3. The failure of responsible players to stop publishing model weights that allow the most dangerous capabilities to propagate freely highlights the need for stronger AI risk management and an AI governance framework.

The tension here is not abstract. It is structural. Responsible actors who self-restrain are simply disadvantaged.

As Mallaby's book makes clear, Hassabis makes many comparisons to Oppenheimer, the scientist who built the atom bomb, then spent his life trying to contain it. The inventors believe they control the technology. Often, the technology controls them.

"Responsible players cannot protect society by doing the right thing alone because the economics of this race punish restraint."

What This Means for Leaders Running Real Businesses in an AI Adoption Era

There is a version of this conversation that happens only in the halls of Oxford, Google DeepMind, and the halls of governments. That version is important, but it is not the one most relevant to leaders running businesses here in Australia.

The version relevant to you is this:

Your ERP systems, WMS, or planning system is not just a record-keeping tool anymore. It is the substrate on which AI in ERP will eventually run. If the data inside it is dirty, inconsistent, or siloed, the AI built on top of it will be confidently wrong, and unlike a human analyst, it will not look uncertain when it is, directly impacting AI decision making.

The "let's wait and see" position is becoming structurally costly. Mallaby's research makes clear that the teams that learned how to work with machine learning in business early, not using it as a shortcut, but as a genuine collaborator, developed compounding advantages. The lag between early adopters and late movers is not closing; it is widening, reinforcing the urgency of AI adoption.

The governance question is yours, not just Silicon Valley's. When your team uses AI to generate a procurement forecast, a store roster, or a customer segmentation model, who is checking the logic? Who owns the decision? If the answer is ‘the AI’ you have an Oppenheimer problem.

The Book's Deeper Argument

Mallaby's sharpest observation is not about technology; it is about temperament. Hassabis believes in depth, in pattern, in the possibility that enough thought and computation can unlock nature's secrets.

He distrusts noise, vulgarity, and sloppy thinking. He also repeatedly underestimates the messier social dimensions of the world he is transforming.

That is not a flaw unique to Hassabis. It is the flaw of every technically brilliant leader who mistakes solving the technical problem for solving the human one.

The protein folding problem was elegant. The governance of a technology that can be weaponised and has already been used in military applications Hassabis did not design for, is not.

The same pattern appears in enterprise technology projects. The system works in testing. The data migration passes UAT. The training sessions run.

And then six months post go-live, the business is working around the system because the humans were never actually brought along.

If you are leading a systems transformation in 2026 and want to think through what AI-readiness actually looks like in your operational environment, not the marketing version, the real one, book a conversation with the 6R team. We work shoulder to shoulder with business and project leaders through exactly this kind of inflection point.

Retail improvement, made practical.
Leadership thinking that drives change.

Sign up to receive new articles and strategic guidance.