02 Purpose 8 min read
From Assigned Role to Chosen Stake
Scarcity used to assign us our purpose. When machines absorb what must be done, purpose becomes a burden of choice — and most people are not trained to carry it.
Meaning looks backward: it is the story that makes what happened cohere. Purpose looks forward: it is the thing you are pointed at. They are constantly confused, and the confusion matters, because automation affects them differently. Meaning has to be constructed. Purpose has to be selected — and for almost all of human history, nobody had to select it.
Necessity used to do the choosing
Purpose was assigned by scarcity. The crop needed planting; you planted. The child was ill; you sat up. The village needed a roof, a smith, a scribe, a soldier. The question “what should I devote my life to?” is a luxury good, and like most luxury goods it was, until recently, available to a tiny minority — aristocrats, monastics, and a handful of scholars who found the freedom bewildering enough to write books about it.
Industrial economies expanded that minority enormously but kept the assignment mechanism intact: the labour market told you what was needed and paid you to supply it. You could choose among assignments. You rarely had to invent one.
Machine cognition attacks the assignment mechanism directly. Not because there is nothing left to do — there is a great deal left to do — but because the things that obviously, urgently, undeniably need you specifically are getting rarer. Purpose stops being issued and starts having to be authored.
Why “find your passion” fails
The advice fails for a structural reason. Passion is usually an output of sustained engagement, not an input to it. People who love their work overwhelmingly report that the love arrived after competence, not before it. Telling an unassigned person to search their feelings for a pre-existing passion sends them looking for something that is generated by the very process they have not started.
Worse, in an era of infinite optionality, searching is nearly costless and therefore never terminates. You can sample forever. The machine will happily help you sample forever — it will generate you a hundred plausible life directions before lunch. Optionality feels like wealth and behaves like debt.
The stake test
A useful substitute for passion-hunting is a test with three questions. A candidate purpose has to pass all three.
- Consequence. If this goes badly, do I feel it — in my time, money, standing or relationships? Purpose without downside is a hobby with good branding.
- Particularity. Does it need me, or would any competent agent do? You do not have to be irreplaceable. But if a model could hold the stake as well as you can, the stake is not doing its job.
- Duration. Can I still be at this in three years? Purposes that resolve in a weekend are projects. Projects are good. They are not purposes.
Notice that none of the three mention enjoyment. Enjoyment is a downstream effect and an unreliable compass; consequence, particularity and duration are things you can check on a Tuesday.
Stewardship as the default shape
The most robust form of purpose available to most people is not invention or achievement but stewardship: taking responsibility for something that would degrade without attention. A body of work. A piece of land. An institution. A craft tradition. A person. A body of knowledge. A neighbourhood.
Stewardship is robust because it is defined by an obligation rather than an outcome, which means it cannot be completed and therefore cannot leave you stranded afterwards. It also has an unusual property in the AI context: the machine can help you steward enormously well, and doing so does not dilute the purpose at all. Nobody feels their care for an ageing parent is diminished by using a calendar.
This is the real difficulty, and the answer is not a better argument but a better architecture. Chosen commitments hold when they are made structurally costly to abandon: made publicly, entangled with other people, given a schedule, given money. Almost every durable human institution — marriage, tenure, apprenticeship, monastic vows, partnership agreements — is a technology for converting a choice into something closer to an assignment. We will need to rebuild those technologies for a context where the market no longer supplies them by default.
Where purpose sits on the ladder
- 1
Cede Machine
The machine does it. You never see the work.
- 2
Approve Machine-led
The machine drafts. You review and sign.
- 3
Collaborate Shared
You and the machine work the problem together.
Once the stake is chosen, pursuing it with full machine assistance is not cheating — it is the point. Leverage on a chosen purpose is the best trade available.
- 4
Critique Human-led
You do the work first, unaided. Then ask the machine to attack it.
- 5
Reserve Human
You do it alone, on purpose, knowing help exists.
The selection itself, and the periodic decision to stay, has to be uncoached. A purpose recommended to you is an assignment wearing a costume.
Take your three largest time commitments and score each on consequence, particularity and duration. Most people discover their biggest time sink fails all three, and something they treat as peripheral passes all three.
Take one commitment you keep half-making and make it expensive to leave: tell someone specific, put it on a recurring calendar, pay for it, or partner with a person who will notice your absence.
Name one thing you are the custodian of. Write down what happens to it if you stop. If nothing happens, you have not found it yet.
Decide in advance how long you will explore before committing — a season, not a feeling. Infinite sampling is the failure mode the tools now actively encourage.