We Know Algos Can Play Us. So Why Not For The Better?
Every scroll, every notification, every autoplay episode is the product of systems that learned our psychological weak points and optimised for them. Addiction, outrage, comparison, FOMO. The algorithms got extraordinarily good at exploiting the patterns that keep us engaged, which is a polite word for trapped.
But here’s the thing. If these systems are sophisticated enough to find and exploit our worst impulses, they’re sophisticated enough to find and nurture our best ones. The capability is neutral. The direction is a choice.
Take something as simple and as profound as friendship.
My closest friendships weren’t built through some mystical process. They were built through maths. Hundreds of small interactions accumulated over time. Touchline conversations at Saturday football, week after week, month after month. College friends forged through years of late nights working through problems together. Slowly revealing whether someone’s humour lands the same way yours does. Whether curiosity runs in the same direction. Whether silence feels comfortable or awkward.
That’s pattern recognition. And we’re building systems that are extraordinarily good at pattern recognition.
I’m not talking about replacing the messy human work of actually building a friendship. That still takes time, vulnerability, showing up when it’s raining and the conversation is just about the ref’s terrible offside call. You can’t optimise intimacy.
But think about the waste. How many potential great friendships have you missed simply because the exposure never happened? You were never on the same touchline. Never in the same room long enough. The bottleneck was never compatibility. It was encounter.
Dating apps tried something adjacent and mostly failed, because they optimised for engagement with the platform rather than the quality of what happens after you leave it. Classic wasted leverage. Extraordinary technology pointed at the wrong target.
Now imagine systems trained not to keep you hooked but to spot genuine compatibility. The subtle signals that predict real connection. Complementary humour. Shared curiosity. The kind of overlap that makes two people light up in each other’s company. Not manufacturing intimacy, just dramatically improving the odds of it having somewhere to begin.
And friendship is just one example.
What about learning? Not optimising for test scores but recognising the specific conditions under which a particular person actually absorbs and retains and gets excited by new ideas. What about health, not generic wellness nudges but systems that understand your specific patterns of stress, recovery and motivation well enough to intervene at exactly the right moment in exactly the right way? What about civic life, community, creativity, grief, the process of finding purpose after a major life change?
We’ve spent many years letting the most powerful pattern recognition systems ever built serve advertising. That’s the real scandal. Not that the technology is dangerous, but that we’ve been so spectacularly unambitious with it.
The algorithms already know how to shape human behaviour. The question isn’t whether they can. It’s whether we have the imagination to point them at something worth shaping.

