AI, and What the Job Was Never About

Vol. 4, No. 10

What remains, once the tasks are delegated, is the part of the work we never quite knew how to describe.

Do we even know what most jobs are about? We talk about roles as if they’re tidy little packages, with defined responsibilities and measurable outputs. In a steady environment, that story holds up reasonably well. But that doesn’t mean it was ever quite true. It just means the gaps didn’t get in the way often enough to raise suspicion.

AI has a way of shining a light into those gaps. If parts of a job can be delegated to agents, what remains becomes easier to notice. Not a neat checklist, but a stream of judgments. What deserves attention, what can be left alone, what “good” looks like when the situation doesn’t quite repeat. The work shifts, almost under the radar, from doing to deciding how to do. That turns out to be harder to pin down.

This is where the idea of a job to be done becomes more situational. The job isn’t sitting there waiting to be described. It takes shape in the moment, under specific conditions, with whatever constraints happen to be in play. From a distance, it looks like a role. Up close, it’s a bundle of activities loosely held together by context.

So when we ask which jobs can be automated, we might be working with an abstraction that’s a bit too clean. It assumes the job is already understood and ready to be broken apart. In practice, it only becomes clearer once you look at it in context. Remove that context, and the work doesn’t fall into neat pieces. It starts to come apart in less helpful ways.

The more interesting question is what happens when execution becomes cheap and abundant. Not which tasks disappear, but which parts of the work were doing more than we had accounted for. Much of what people do never made it into systems or descriptions in the first place. It lives in how situations are read and handled over time. Once you start looking at work through that lens, it becomes harder to claim we ever fully understood what the job was in the first place.

The Factory Fantasy

AI changes the economics of parts of software development. That much is obvious. It does not erase ambiguity. It does not replace judgment. It does not solve coordination. It does not remove the need for accountability. And it definitely does not turn engineering into a factory just because a consultancy found a metaphor executives like. That is the part builders can see immediately. That is also the part McKinsey keeps missing.

Ilya | 10 Minutes

Automation‘s appetite for human traces

These two ingredients, context and judgment, are both crucial elements for labs and companies wishing to use agents to replicate human work. To automate a role, you need to clearly define what it is and how to tell when it has been done well. But for many roles, like the one of the scientist, the philosopher, or the statesman, the criteria for success are fuzzy, the effects of your work show up on long time horizons, and the impact of any given decision you take is hard to trace.

Hamidah Oderinwale | 14 Minutes

The Multiplier Isn’t AI. It’s Where You Point It

AI’s return depends enormously on where you apply it, what you’re asking of it, and who’s at the keyboard. That’s the individual-level version of the multiplier question. The version I’ve been rolling around since Mike spoke is the organisational one. Because whether AI amplifies individuals at different rates matters less, I think, than where organisations are actually choosing to point it in the first place.

Lisa Woodall | 9 Minutes

What jobs are AI jobs?

And the thing that I was sort of thinking about as maybe getting to a high level of abstraction here, is to say, you know, okay, Christensen, phrase, what is the job to be done? What is the actual thing that the customer is buying from you? And how do you map that against what is the thing that this new technology is changing? And so you can look at, and where is the point of leverage? We’re realising we actually don’t know what the job of most people is to do.

Benedict and Toni in Another Podcast | 30 Minutes