AI & Work: Why the Future Might be Weirdly Human

I’ve written a few posts on AI.

Last week’s was on what the transport sector can teach humanity about what’s coming, based on our history. You can read that here, if you’d like to. In fact, I rather hope you will - I’m pleased with this one.

Often, when I write on AI, I’m assailed by social media comments telling me I’m an idiot to be taken in by the hype and nothing’s going to happen.

Then I come onto LinkedIn and see another post explaining that no-one will have a job in five years, and we’ll all be watched over by machines.

They can’t both be right, and no-one knows the truth.

So this isn’t a post about what I think will happen. It’s a post about what might. Think of it like a risk register entry: even if low probability, high impact. The kind of thing you don’t plan the entire business around but you’d be daft to ignore entirely.

Because even in the most extreme scenario - the one where AI eats everything - I don’t think humans are finished.

We’ve been here before. Several times.

When I was at school, we went on a trip to the Ford factory at Dagenham. The production line was pretty much all robots with an occasional human. It was very weird. The humans were clearly the few left after their colleagues had been replaced.

Today, the Dagenham factory’s gone. But even if it hadn’t, the humans would be. And not because anyone wanted them gone: just because machines got better.

The UK still makes as many cars as it did in the 1970s. It just does it with about 10% of the people.

Now apply that to HR. Or finance. Or planning. Or transport scheduling. It’s not hard to see where this is heading.

A lot of white-collar work may be halfway out the door already. Soon “the HR department” may be one senior person managing an AI that does everything from recruitment to policies to payroll. Ditto legal. Ditto much of finance.

The AI won’t be unsupervised but it may do most of the work: a bit like having an army of hyper-bright interns who never sleep and sometimes make catastrophic mistakes.

And this is very much a transport story too.

AI is coming for:

Route planning
Timetable optimisation
Contract drafting
Consultation summaries
Customer complaints triage
Ticketing anomaly detection

… and the rest.

This isn’t science fiction, it’s “just software”, to quote Mustafa Suleyman, the founder of Deep Mind - now Google’s AI Lab.

Trains already drive themselves

Most trains in London now drive themselves. Most passengers ticket themselves. Our grandparents would be completely baffled to see us walking into the station and tapping our watch onto the gate and for it to open the gate and charge our bank account. Yet it does.

Despite that, there are still plenty of humans.

Why?

Because what’s been automated is the task, not the experience.

People value lots of different outcomes. Punctuality, convenience, reassurance and comfort are all things people care about and, while some can be fully automated, others (like empathy when something goes wrong) can’t be.

That’s why people are happy to travel on the Victoria line (which is automated), but want to know there’s a driver on board. The automation provides the punctuality and frequency, but the driver provides the reassurance.

Lets compare two places to buy a coffee:

A McDonald’s drive-thru with a touchscreen and no eye contact.
Your local independent with the nice barista who asks how your day’s going.

Both serve coffee. Only one serves you.

This is an important point. Automated food ordering using touchscreens has been possible for over a decade. So it hasn’t failed to spread because it can’t be done. It’s failed to spread because in certain contexts, we don’t want it.

Work Will be Human - but not as it was

AI may get good at doing certain complex tasks, but it’s not in charge of relationships.

Even if machines become as good at empathy as we are, humans are needed to take responsibility, especially when it matters. It’s like why a Director insists on seeing a Board paper before submission: some decisions are too important to delegate.

The same will be true of AI. However good it gets, someone still needs to check the work and to ask, “Is this the right call?”, especially when that call affects other people. The subordinate to the director is capable of making the call, but the director’s job is to make the call. That will still be true of humans.

Enterprise software will become much more about AIs talking to AIs, running automated workflows between departments. But that doesn’t change the fact that a manager is responsible and has to make the final call. No-one thought that robots were in charge of Dagenham, even if they made the cars.

That means humans will still be needed but in different ways. The skill of briefing an AI will be crucial, as will designing an AI-enabled workflow, understanding where humans need to be in the loop. The skill of shaping corporate systems to maximise the value of AI agents who will still hallucinate and get things wrong will be difficult, though AI can help. We may see more professional ethicists.

Being good at Excel will become less about formulas (which AI can create from prompts) and more about knowing what to measure. That will democratise access to more complex analysis - which means we’ll do more. Almost certainly, much too much. It won’t make us more productive, but it will keep us busy.

Being good at writing will be less about grammar (which AI can do effortlessly) and more about knowing what matters. Strategy will be less about reports and PowerPoints and more about choosing questions and running scenarios.

In transport, this changes how we think about some of the most familiar roles.

Imagine:

A planner who oversees AI-run scheduling, but adjusts based on local knowledge that the school timetable data the machine used doesn’t reflect the fact that the kids at Trinity Green always hang around in the supermarket car park for 15 mins before dispersing.
A scheduler who lets AI generate 100 network simulations, but picks the one that creates a time-off pattern that reflects the local community’s aspirations for leisure, based on their work with the local authority.
A station manager who uses instant AI predictions of the severity of an operational incident to know when to position themselves on the concourse to apologise to customers in person, and when not to worry.

In all these cases, the AI will be a tool but it will not be right. AI kool-aid visions of AI nirvana will not come true. The capabilities will be remarkable but they will not be God-like.

Customer Experience Will Be the New Battleground

Imagine if I told you I’d simply asked ChatGPT to write me a post on the future of work in the context of AI and publish it on this blog, and that was what you were reading now.

Would you value it as much?

No? But that’s mad, surely? Isn’t the value in the words and insight, not in the author?

Well, apparently not.

You're not reading this blog post because you have to, but because you want to (and thank you, by the way!). The fact that it was written by someone you’ve heard of (and, very possibly, personally know) is part of your motivation.

In future, things we have to do can be automated, and will be. But the things we want to do will be more human than ever. The more samey and frictionless everything dull becomes, the more people will crave the unpredictable, the personal, the properly human.

If that’s irrational, well - who knew: humans are irrational.

But it means our current approach to transport automation is slightly bonkers.

We’ve automated many of the hard bits - driving, ticketing, rostering - but often kept the remaining humans out of sight. I believe there’s real value to the person on the train but they have much more value when they make an empathetic announcement if there’s a delay and even more again if they can be seen.

There are many reasons why the DLR has such high customer satisfaction scores but - I believe - a big one is that it’s got some of the most visible human staff despite being the country’s only automated railway serving unstaffed stations.

*My daughter being offered the chance to make an announcement by a brilliant member of DLR staff. This is the future of transport in an AI world.*

That’s the model for the future: as machines take over tasks, people will crave positive, enjoyable human experiences.

You can see this happening on the high street. In posh areas (which, amazingly, now seems to include Walthamstow Village), the supermarkets are still there, but they’re joined by butchers, bakers and greengrocers. The kind of shops that vanished when we all embraced efficiency… but which are now making a comeback. Why? Because people like them. They enjoy the interaction. They’re willing to pay for it.

*An expensive greengrocer has opened on Orford Road in Walthamstow Village. No-one needs a shop like this: but they want the experience.*

Transport needs to master this zeitgeist.

Right now, much of public transport still feels utilitarian. It feels like our goal is to make it efficient and functional, but that often means joyless. And that’s when we get it right.

If we want people to choose public transport - not just use it when they must - we need to make it experiential: warm, personal and human.

The AI future is actually about:

The bus “driver” who greets you. (they may not actually drive the bus)
The member of staff who makes a joke about your suitcase.
The train that feels calm.
The station that offers you a sense of place not just throughput.

That’s the future we need to build: full of humans, just not doing what they do today. The point isn’t that a robot wouldn’t be capable of making a joke about your suitcase, it’s that no-one would laugh.

The real world still matters

One final point it’s worth making when in the business of predictions about AI: the world is still full of politics, trade unions, public opinion and human frailty.

Even if the machines can do a job, that doesn’t mean they’ll be allowed to. (Just ask anyone who’s tried to reduce bus mileage in an election year or reduce booking office opening hours.)

This is one of the big blind spots in the more extreme AI narratives. They imagine a clean-room future: no messy humans, no dissent, no newspaper headlines.

But that’s not the world we live in. The real world is noisy and contradictory and the more we automate, the more we’ll need people to manage what AI doesn’t see: nuance, culture, values, trade-offs.

AI can save public transport — but only if we let it

There’s a lot of hype around autonomous cars.

Let’s be clear: it’s absolutely possible they’ll never work.

Tesla’s Robotaxi rollout in Austin, Texas, is a swizz. An autonomous car isn’t autonomous if it’s being driven on video instead of in the driving seat. That’s just a less safe manual car.

But autonomous cars might happen and if people like me who don’t currently drive start using them, we’ll have gridlock. AI can optimise, but it can’t repeal physics: only one object can occupy one space at a time. Autonomous or not, a car takes up road space, and a lot of it. Autonomous cars risk more cars in already crowded cities. That would not be a good outcome.

But autonomous public transport is different. A driverless shuttle operating within a defined suburb can be backed up by a remote operations room. One human supervising hundreds of vehicles, ready to step in if needed would make suburban, high-frequency, electrified shuttles not just viable but game-changing.

*Ruter Autonomous Vehicle of the type I test-rode in Oslo*

On the back of a fag packet, I estimate that we can run autonomous public transport at £1 per mile. On the podcast, I spoke to Christian Willoch, who’s in charge of the autonomous vehicle programme at Ruter, the public transport authority for Oslo and the person leading the most advanced project in Europe. He has a much better fag packet than me, and he endorsed that £1 per mile figure.

At that level, the entire economics of connectivity flips. Towns, villages, new suburbs that are currently cut off from public transport all suddenly become within reach. And as the peripheral network grows, the core network strengthens too. It’s the opposite of the Beeching Cuts. That will… add jobs.

Now, you may easily point out an inconsistency here. Earlier I said that people want a human on the Victoria line but I’m willing to bet on people being willing to use autonomous shuttles. You’re right. I can’t be sure they will be willing to do so. But once upon a time, people demanded lift operators and now they don’t. So my judgement is that if a small, self-contained vehicle feels safe, people will be willing to travel on it without staff. It’s very different to the tube.

I’ll be honest: I wish we were moving faster on this. We didn’t make as much progress as I’d hoped at TfL. As I said at the start, this is low probability but exceptionally high impact. I really hope every public transport authority has someone in charge of working up the opportunities from autonomous public transport. The opportunity is real. But we’ve got to want it. Happy to chat further with any public transport authority interested in doing something in this space.

We’ve been here before. We’ll go through it again.

We’ve already lived through multiple automation revolutions. And every time, humans found new ways to create value. The danger isn’t that we’ll be replaced, it’s that (in transport!) we won’t move fast enough.

Given my post last week (do feel free to read it!) is very negative about the AI companies, you might be surprised I’m saying AI is going to have so much impact. That’s because me not liking Sam Altman doesn’t mean that Sam Altman’s company isn’t going to be influential. AI is happening, whether you want it to or not.

As someone who freely acknowledges that all my predictions in this space will be wrong (so will yours. So will everyone’s), my view is that the challenge for transport is as follows:

Automate as many functions as possible
Humanise the experience as much as possible

Together.

Now I’ll press publish and wait for everyone to call me an idiot…

I’m going to come back to this topic in future. Subscribe for more.

I help organisations get serious about innovation and future change.

If this is what you need, let’s talk.