the engine, not the driver

A week ago I wrote about where AI is taking the economy. Short version: it will run it, better than we ever did. The AI agents will buy the shoes, match supply to demand, take over the small daily decisions – and remove the moment where we used to stop and think “do I actually need this.

It can do all of that. What it can’t do is decide what all of it is for. That it has to ask us:

optimize for what?

This post is where I try to take that question further.

Here’s the obvious answer, the one I reached for first. If the machine will optimize whatever we tell it to, then we just need to tell it better things. Stop feeding it ‘cheapest’ and ‘more.’ Feed it ‘wellbeing,’ ‘sustainability,’ ‘a good life.’ Give it better goals and let it optimize toward those instead.

It sounds right. My AI system kept gently complicating it. And seeing why it’s wrong is the whole point.

Because the moment you hand a machine “a good life” to optimize toward, you’ve already turned a good life into a target – a number to climb, an output to maximise. You’ve kept doing exactly what the economy always did: treat living as something to be produced efficiently. You just swapped the metric.

The machine doesn’t free you from that logic. It perfects it, faster, with prettier words on the dashboard. Quantifying life is classical thinking in a machine version.

And there’s a quieter problem underneath. If the machine optimizes the good life for you, then you’re not living it. You’re receiving it. The deciding – what’s enough, what matters, what to choose – has been handed off again, the way we handed it to the economy a long time ago. Different master, same outsourcing.

The problem was never that we gave the machine the wrong instructions. It’s that we keep looking for someone – something – to do our deciding for us.

For a few centuries, that something was the economy. Ask it what a good life is, and it always had the same answer ready: more. The scoreboard answered, and we stopped asking, because we didn’t have to. It was convenient. It worked, sort of. And it quietly trained us out of a habit humans used to have: deciding, for ourselves, what a life is for.

The machine is now offering to inherit that job. And the temptation is enormous, because it’s so good at it. But the answer to “optimize for what?” isn’t a better optimization. It’s to stop optimizing the question away. To take the deciding back – not give it to the economy, not give it to the machine – and aim it at something the economy never could: a good life chosen, not produced

That’s the real shift. Not a better goal for the machine. A different kind of goal, held by us. The economy asks: how do we get more, faster? A life asks something a machine can’t be pointed at: what do we want to be? What’s enough? How do we want to live, and with whom? You don’t reach those by optimizing. You reach them by choosing. By living one way and not another. They’re life-bound, not economy-bound – and they were never the machine’s to answer.

This is about values. About staying human while the machines get better at being us.

This is what we’ve been working on at the Business Hippie Club, though it took the AI question to make me see it this clearly. We’ve been building something called the 17 Living Goals. Seventeen plain attempts to name what a good life actually asks for – across four areas: how we live, how we work, how we feel, what we give back. Not instructions for a machine. Questions for people to decide by – in a life decision, a business choice, a city plan.

Two of them, as examples.

One — living goal 04 — is “enough instead of endless.” Notice you can’t optimize toward “enough.” Optimization only knows endless; “enough” is the point where you choose to stop, and a machine never chooses to stop. It’s a human decision or it’s nothing. The economy doesn’t have the word because the economy never stops. We’d have to.

Another — living goal 13 — is “belonging in an age of disconnection.” Try handing that to an optimizer. There’s no metric to climb, no output to maximise. Belonging isn’t produced; it’s built, slowly, by people choosing each other. A whole category of what makes life worth living that no scoreboard and no machine can deliver – only we can.

That’s why these are goals and not instructions. You don’t feed them to a system. You decide by them.

I want to be careful here, because this is where things usually go wrong. The 17 Living Goals are not a finished system. They’re a direction – a working draft for a kind of deciding we’d mostly stopped doing. Some of the goals are probably wrong. Some are missing. The point isn’t getting the list right. The point is to start asking the questions again – ourselves, out loud. Before the machine answers them for us, by default.

None of this means refusing the machine. It’s extraordinary – let it carry the logistics, the routes, the drudgery, the thousand things that really are input-output problems. Let it reason and suggest. But the goal it serves has to be ours, and the right kind of goal: quality of life, nothing more.

However powerful the engine, it was never meant to be the driver. and the seat doesn’t stay empty. The machine is already running, on the only answer anyone ever gave it: more. Every day we don’t take the wheel, that answer hardens into the default. The deciding we don’t do, something else keeps doing for us – this time without ever pausing in the shop to wonder if we actually need this.

So this is less a conclusion than an invitation. We have to start deciding again what a good life is – and not leave it to the scoreboard, or the machine. I’ve made a start with developing the 17 Living Goals. It’s unfinished, on purpose. Tell me where it’s wrong.