The Jetsons Didn’t Prepare Me for This

 

Six possible futures for work, and one fairly confused human.

Can I even return to writing without talking about the obvious first?

I’ve spent a fair amount of time lately debating how our jobs will evolve, what will survive, and what we will all do if AI eventually becomes capable of doing most things for us. Sometimes this feels like an interesting intellectual exercise. Sometimes I remember that I have a career and bills, and the exercise becomes considerably less intellectual.

Yesterday, a conversation with a couple of friends took us through several versions of this future. We didn’t solve anything, obviously. But we did move beyond the usual “AI will replace you” versus “someone using AI will replace you” debate.

So, before anyone calls this AI slop, the existential confusion was very much a group effort.

I can see at least six directions this could take. I’m sure there are more, and I’m increasingly doubtful that we get to pick just one.

The first is the one that makes corporate leaders excited and everyone else update their CV: companies learn to operate with a fraction of their current workforce.

Imagine an organization doing roughly the same work with a third of its headcount. Perhaps less. There would still be people making decisions, owning outcomes and dealing with whatever the machines get wrong. But far fewer of them.

I can understand why an individual company would pursue that. If it can deliver the same quality, faster, at lower cost, the incentive is fairly obvious.

What I struggle with is what happens when everyone succeeds.

Employees are also consumers. The people whose salaries disappear are the people buying subscriptions, booking holidays, upgrading phones and paying for the things these businesses sell. At what point does reducing the cost of producing everything start reducing the number of people who can afford anything?

I don’t think that contradiction prevents companies from cutting jobs. Individual businesses can make perfectly understandable decisions that create a rather difficult collective outcome. It does make me wonder how stable this version of the future would be.

The second possibility is that the economics don’t work quite the way we expect.

We tend to discuss AI capability as though access to it will be unlimited and increasingly affordable. Perhaps it will. But what if the most capable systems remain expensive, especially when we ask them to work continuously, solve difficult problems and produce reliably good results?

The relevant comparison would be the full cost of getting useful work done. That includes checking the output, fixing mistakes, protecting information and having someone answer for the consequences.

In that world, businesses might use cheaper AI to help people with everyday work and reserve expensive systems for tasks where the returns justify them. Humans would remain part of the equation because, for some work, we would still make economic sense.

There is something mildly humbling about imagining that the thing protecting our jobs might be the size of someone’s AI invoice.

The third possibility is the reassuring one: new jobs will emerge.

I find this plausible. I also find it difficult to use as reassurance for someone worried about their next five years.

We may create professions that sound absurd today. We may discover entirely new services people want once other things become cheaper or easier. We may spend more on experiences, care, creativity or human attention.

But the practical questions remain. How many jobs? Where? Requiring what skills? Paying how much? And how long is the gap between someone losing their existing livelihood and being able to earn through a new one?

Working in HR makes that gap particularly difficult to ignore. A new category of employment appearing somewhere in the economy does not automatically give a displaced employee a realistic path into it. They may need training, money, time or the ability to relocate. They may have none of those.

I can believe that new work will emerge and still worry about the journey there.

The fourth possibility is that the most powerful AI never becomes something all of us can casually access.

I’m imagining systems capable of making major scientific advances or tackling problems on the scale of the Navier–Stokes existence and smoothness problem. That isn’t a definition of AGI; it’s the level of ambition I have in mind.

Would capabilities like that be available to anyone with a monthly subscription?

Perhaps. Or perhaps access would be concentrated among governments, major research institutions and a handful of companies. Defence, drug discovery, infrastructure and sophisticated trading would all give organizations reasons to pursue an advantage and reasons to keep it.

Then add governments deciding what can be deployed, companies protecting their intellectual property, and countries taking different positions on acceptable use.

We might end up with very different AI futures depending on where we live and who we work for. Everyday assistants for most people; something substantially more powerful behind a much more expensive or restricted door.

The fifth possibility is where the conversation becomes a film.

We decide things have gone too far and pull the plug.

The Skynet version is easy to picture. The less cinematic version could involve systems making consequential decisions that people cannot reliably understand, constrain or reverse. They wouldn’t need to be conscious for that to become a serious problem.

But even the phrase “pull the plug” assumes a surprising amount.

Who gets to decide? Would competitors stop together? Would countries? How willing would we be to switch off something once essential services and large parts of the economy depended on it?

I’m not predicting this outcome. I’m questioning how confidently we assume that stopping will remain a simple option.

The sixth possibility is the one that sounds closest to the future we were promised as children.

Machines do most of the work. People receive some form of universal income or share of the resulting wealth. Employment becomes optional, or at least much less central to survival.

There is a lot to like about that idea. More time with people we care about. More room to learn, create, travel or simply exist without having to justify every hour economically.

Then my questions start again.

Who owns the machines? How do their gains reach everyone else? What can poorer countries afford to provide? How do we reward difficult work that still needs a person? And what happens to status and identity when a job title stops doing so much of the work of introducing us?

Many of us would welcome freedom from parts of our jobs. We might still miss being needed, being useful, having a routine, or belonging to a group working towards something.

How countries would fund this and encourage contribution, I honestly have no clue. I suspect the political and social arrangements would be at least as difficult to invent as the technology.

After going through all six, I keep returning to a seventh possibility: we get bits of all of them.

Some organizations shrink dramatically. Others discover that automation is harder or more expensive than the demonstration suggested. New professions appear while older ones contract. Certain capabilities become commonplace; others remain tightly controlled. Governments intervene at different points, for different reasons.

That would leave us with an uneven transition in which two people can describe completely different experiences of AI and both be telling the truth.

I watched The Jetsons. I watched Back to the Future. I was prepared for flying cars and questionable fashion choices. Neither prepared me for wondering whether my career would outlast my home loan.

And yet, I’m curious enough to keep experimenting.

For now, I can learn the tools, test them against actual work, notice where they help and where they fail, and keep revisiting my assumptions. In my professional life, I can also ask better questions about what happens to people when we redesign work. The efficiency calculation deserves a serious look at the transition it creates.

I don’t have a confident forecast to end with. I would quite like to hear yours. Which of these futures feels most plausible? What have I missed? And what are you doing differently today because of it?

There’s a line from The Shawshank Redemption that feels appropriate:

“Hope is a good thing, maybe the best of things.”

I’m keeping the hope. I’m also learning the tools.

Tomorrow still has work to be done.