TL;DR

  • Modern knowledge work has quietly built identity on producing things - and AI pressure makes that fragility visible without you having to lose your job to feel it
  • History (Keynes’ 1930 prediction) suggests freed-up capacity defaults to “more work”, not leisure - the shift to meaningful work has to be chosen deliberately
  • What stays valuable when execution gets cheap: deciding what is worth doing, taking responsibility, sitting with other humans, craft for its own sake, and growing other people
  • The “everyone will do deeper work” narrative ignores the dignity problem - for many people, work is structure and belonging, not just a vehicle for meaning
  • Put your meaning somewhere that does not depend on being the cheapest producer of an artefact - it was never a secure place to put it, and agents are just making that clearer

This is another “thinking out loud” post, in the same spirit as the agent-first architecture piece. I do not know how any of this is going to land. I am writing it partly because the question has been rattling around in my head for months, and partly because I suspect a lot of people in and around software are quietly wondering the same thing without quite wanting to say it out loud.

The question is roughly this. If the optimistic version of the AI trajectory plays out - if agents really do absorb large chunks of what we currently call “the work” - then what is work actually for? Not economically. I mean for the person doing it. What is left of meaning, identity, and self-respect when the doing itself has been handed off?

I do not have a clean answer. I have some threads I keep tugging on.

A Note on the Word “Abundance”

“Age of abundance” is a phrase that gets thrown around a lot in AI-adjacent writing, and I want to be careful with it. It is easy to treat it as a done deal. It is not.

What I mean by it here is narrower and more specific. I mean a plausible near-future in which the cost of producing a lot of useful digital work - code, analysis, writing, design, research, coordination - drops sharply, and where the bottleneck on getting useful things done stops being “can we find someone to do this” and starts being “should we do this, and what exactly do we want”.

That is not post-scarcity. It does not dissolve rent or food or healthcare or the need to pay for any of it. It is a much more specific shift: the disappearance of a particular kind of friction, in a particular kind of work, for a particular slice of the economy. If that shift happens at the scale people are predicting, I think it forces a conversation about meaning that the people in the affected slice have largely been able to avoid.

That slice includes me. So I want to think about it honestly.

Where We Currently Put Meaning

It is worth saying out loud what most of us actually do with meaning, because the default is usually invisible.

A lot of modern knowledge work quietly assumes that producing things is the meaning. You are what you ship. Your value is legible in the artefacts - the PR, the report, the talk, the launch, the quarterly number. Derek Thompson called the belief system around this “workism”, which I think captures it neatly: the idea that your job is not just a job but the primary source of identity, community, and self-worth.

It is a fragile place to put meaning even in good times. It breaks when you get made redundant, when you burn out, when a reorg shuffles you into a role you do not recognise. What AI pressure does, I think, is bring that fragility forward and make it more visible. You do not have to lose your job to feel it. You only have to watch an agent one-shot a task you used to be proud of.

Hannah Arendt, in The Human Condition, drew a distinction between labour (the repetitive stuff that keeps the biological machine going), work (the production of durable things), and action (doing and saying things among other people that leave a mark). Most of what AI is visibly absorbing first sits in her “labour” bucket and the lower end of “work”. That leaves “action” mostly alone - but action, in her sense, is also the hardest to measure, the hardest to turn into a job title, and the hardest to anchor an identity to, because it depends on other people as much as on you.

If you built your identity on labour and work, and the labour and work layer starts to compress, you are going to feel it. That is not a failure of character. It is just what happens when the ground you were standing on moves.

The Grandchildren Problem

John Maynard Keynes wrote an essay in 1930 called “Economic Possibilities for our Grandchildren” in which he predicted that by around 2030 the economic problem - how to feed, clothe, and house people - would be mostly solved, and the grandchildren in question would work perhaps fifteen hours a week. He was roughly right about productivity and roughly wrong about the fifteen hours.

Almost a century later, we are much richer in aggregate and we work about the same. Keynes was writing about physical abundance. He did not anticipate that the demand for status, novelty, and positional goods would expand to absorb the slack. He did not anticipate that work itself would become a status good.

I find this history useful as a caution. The natural assumption is “if agents do the work, we will work less and live more”. History says: probably not, at least not automatically. The slack tends to get filled. If anything, the people whose work an agent did not fully absorb tend to end up doing more, not less - because now they are responsible for supervising the agent as well as doing the residual human part.

So when I catch myself writing a sentence like “agents will free us up for deeper work”, I try to pause and ask what the honest version of that sentence is. The honest version is probably: “agents will free up capacity, and what fills that capacity is a choice - one that defaults to ‘more’, unless we do something different on purpose”.

What Actually Stays Meaningful

Here is where I have to make the switch from diagnosis to something more useful, because “everything is fragile” is not a post.

The threads I keep coming back to, when I ask what is likely to stay meaningful once the easy production is cheap, are roughly these.

Deciding what is worth doing. Agents do not ask, unprompted, whether something should exist. They answer the questions you bring them. The act of choosing the problem - the taste, the judgement, the willingness to say “not this, this other thing” - seems to become more valuable the cheaper execution gets, not less. It is also one of the parts of the job least well captured by a spec.

Taking responsibility. When an agent fleet does something wrong, the fleet did not do it. A person did, somewhere upstream, when they chose the objective, the guardrails, and the definition of “good enough”. Accountability seems to concentrate rather than dilute in an agent-heavy world. The person willing to hold the bag is still rare, still needed, and still - if you have ever worked with one - noticeably different in quality from the person who is not.

Sitting with another human. The slowest, most expensive, least scalable activity in knowledge work is also, in my experience, the one most often responsible for the thing that ends up mattering: an hour with a stakeholder where you actually understand what they need and why. Agents can help prepare for that hour. I do not think they can replace the hour. If they can, I suspect we have bigger problems than meaning.

Craft for its own sake. There is a quiet argument that says the only reason to do careful work is productivity, and if the agent is more productive, there is no reason to do it yourself. I do not buy this. People still play chess against humans after engines surpassed us. People still do long-form arithmetic to learn maths even though calculators are free. Flow - the state where challenge and skill meet - does not require your activity to be economically load-bearing. It just requires it to be hard enough, and to matter to you. Craft as a source of meaning survives automation. Craft as the only source of meaning probably does not.

Teaching, mentoring, gardening other people’s growth. This is the one I am most confident about. The value of a senior person who makes three junior people 20% better has always been underweighted, because it is hard to attribute in a PR description. In an agent-heavy world, it gets more important, not less, because the humans around you are going to be making bigger decisions on less experience, with faster feedback loops, and they need someone to help them calibrate.

None of these go away when execution gets cheap. If anything, they become more of what the job is.

The Dignity Problem

There is a version of this conversation that skips the hardest part, so I want to name it.

For a lot of people, work is not primarily meaning. It is dignity. It is the thing that makes you a full participant in the world - income, yes, but also having somewhere to go, something to show up for, a group of people who notice whether you turned up. The classical philosopher’s answer to “what will people do in an age of abundance” tends to assume everyone finds craft or leisure or civic life naturally satisfying. In practice, a lot of people find structure, routine, and belonging through the job - and those are not trivially replaceable with a universal basic income and a hobby.

I do not have a clean view on the macro version of this. UBI, shorter weeks, different safety nets - those are above my pay grade, and the people who claim certainty about them tend to be wrong in predictable ways. What I do think is that any honest “meaning of work” piece has to at least register that the loss of routine work hurts people in ways that “you will be freed up for higher-order thinking” completely ignores.

If you are someone who still has agency over your own working life - most people reading a blog post like this are - I think you have a quiet obligation to notice the asymmetry. Not to solve it. Just not to pretend it is not there.

A Personal Version

Let me try to write the honest personal version of this, because the abstract version is easy to retreat into.

I have spent most of my career being rewarded for producing things. Pipelines that ran. Systems that held up. Deliverables. It is flattering to the kind of person who likes a clean commit graph and a visible output. It is also, I now think, a somewhat impoverished definition of a working life.

When I imagine agents doing more of the producing, my first instinct is a kind of low-level panic, because my identity has been quietly entangled with the production the whole time. When I sit with it longer, the panic turns into something more like curiosity. If I am not the person who produces the thing, what am I? And the answer that keeps coming back is not “nothing”. It is “someone who decides what the thing should be, who it should be for, and who carries it when it goes wrong”.

That is a harder job, not an easier one. It is also a more interesting one. And it is closer to what I have actually enjoyed about the best bits of my career than the ticket-closing ever was.

I do not want to romanticise this. There is no guarantee the economy rewards the hard, judgement-heavy version of work as generously as it rewarded the production version. There is no guarantee the companies I might work for will want me to do it. There is no guarantee I will be any good at it. But as a personal working hypothesis about where to put my energy, it feels less hollow than “get faster at shipping so I can keep up with the agents”.

Habits I Am Trying to Build

This is the part where a “future of work” post usually offers a clean five-bullet framework. I do not have one. I have a few habits I am experimenting with, and I will probably drop half of them in a year.

  • Write the intent down. Before I pick up a tool - my own hands or an agent - I try to write a paragraph on what I am actually trying to achieve and why. The discipline of specifying turns out to be the discipline of thinking, and it is the part of the work that does not get cheaper when execution does.
  • Name the thing I am not going to automate. Even if I could. Some activities - reading code carefully, talking to a stakeholder, writing a first draft by hand - I am deliberately keeping manual, because I think the slow version is where I actually learn, and I do not want to trade the learning for speed.
  • Invest in people. Mentor, pair, give real feedback. This has always been leverage; I think it is more so now.
  • Keep a small portfolio of useless work. Music, writing like this, side projects that will never ship. Not as a hedge - as a reminder that the part of me that values making things is not contingent on any particular employer rewarding it.
  • Read outside the field. The people who are best at thinking about work in an age of automation tend not to be the ones building the automation. Economists, historians, novelists, philosophers. They have been through this before, in different clothes.

None of this is advice. It is just what I am doing while the ground moves.

A Note on Fear, Again

I wrote a version of this note at the end of the agent-first architecture piece, and I am going to repeat it here because it matters.

I am not telling you jobs are safe, or that “everyone will just do deeper work now”. I have no insight into that, and the people who claim to usually turn out to be selling something. What I am saying is that the question “what is work for” is worth sitting with directly, before the answer gets handed to you by a product roadmap or a layoff email.

If the optimistic version of the next decade plays out, a lot of the grinding, low-leverage, repetitive parts of the job become cheap, and the interesting parts - the judgement, the taste, the responsibility, the care - become more of what you actually do. That is the version worth aiming at.

If the cynical version plays out, a lot of the same grinding parts become cheap, get handed back to us in a different shape, and the interesting parts stay scarce, unevenly distributed, and under-rewarded. That is the version worth preparing for.

Either way, the advice ends up in the same place: put your meaning somewhere that does not depend on being the cheapest producer of an artefact. It was never a great place to put it anyway. The agents are just forcing the point.

If you want more of me thinking out loud about where this might be going, the last few posts in the AI section are all orbiting the same question from different angles. None of them are predictions. They are all me trying to figure out, in public, what any of this is for.

I would rather be usefully wrong about it than quietly anxious.