I can’t rinse the dishes until the flow rate of my kitchen tap is Pareto-optimal.

A gush wastes water; a trickle wastes time. Tinkering with the tap is a meta-problem that wastes both.

Inefficiencies gnaw at me like a rusty bike chain. Processes must be optimised, smoothed. My dad is a data analyst, as are both his brothers. Maybe it’s genetic.

But whether it’s busywork screaming to be scripted, the 2:1 in my bachelor’s, or crossing the road along anything other than the hypotenuse: if performance can be analysed, it can be optimised. And this new era leaves plenty of room for optimisation.

I remember the day I had my first conversation with a machine. GPT-3.5, specifically. I knew something about the world had fundamentally changed.

The preeminent 20th-century philosopher, Wittgenstein, said that the limits of our language are the limits of what we can know. “If neural networks modelled game-playing better than humans,” I thought, “and they can now model language, then…”

My mind raced with implications. How much potential could this unlock for the economy? For society? And how much is just waiting to be discovered?

That day, I knew I had to do two things: study the fundamentals of AI, and get the top grade whilst doing so.

I did both during my Master’s, working with the NHS and Oxford along the way.

Then an engineering multinational hired me. I saw three hundred staff inspecting buildings with pen and paper, emailing themselves photos, and typing up reports back at the office. I built an app that writes the report in one click. That saved 10,000 hours, before any AI.

What could I optimise for you?

Ben Grime