How A.I. Can Benefit Transport, Part 2
One of the challenges I always face when writing about AI is that the world seems to divide between the tech bros, who believe it’ll cure cancer (before killing us all), and the haters, who believe data centres are divisive and large language models are climate catastrophes.
I think it’s likely that all of these perspectives are true.
Whether or not it cures cancer (or kills us all), it’s an incredibly potent technology that will have a huge impact. It is also a carbon disaster that depends on vast data infrastructure.
But the key point is that it’s happening regardless.
So if AI is happening, how can transport benefit as opposed to being left behind?
Last week, I offered up three ideas:
We can transform our response to disruption
Customers can access live, dynamic journey planning
Safety can be improved by the use of tech with judgement
If you have time, do read that post before this one.
Today, I’m going to look at some other ways in which AI can benefit our organisations.
Iteration and prototyping
Early next year, I’ll be publishing the outputs of a project I’ve been working on with colleagues in Oslo, Beta Mobility.
As you probably know, I do a lot of work in the field of integrated transport. It frustrated me that we don’t have a recognised measure for how good a city’s integrated transport actually is. I think London’s is better than Liverpool’s and that Munich’s is better than Manchester’s. But that’s just my opinion. What if there were a recognised measure, from which we could create a league table?
Given that there isn’t one, I decided to create one and asked my Norwegian colleagues to help.
Why did I ask Beta Mobility? Because several years ago, they predicted the way the world was going and completely retooled themselves around AI. I wrote down my ideas for what this index could look like and how it could work.
24 hours later, it existed.
Now, obviously, it wasn’t perfect (it still isn’t - that’s why you’ve not seen it yet).
But the point is that Johan (he’s the chap who runs Beta Mobility) didn’t spend days (let alone weeks or months) writing a specification. All our effort has been focused on iterating something that actually exists in real life.
That’s how life will be for all of us.
At the moment, it’s still largely specialists like Beta who work that way, but it won’t stay like that.
Eventually, AI will democratise access to software and prototyping in the way that Excel has democratised access to financial modelling.
When my mum was a manager at the BBC back in the 1980s, financial information was produced by Accounts. A normal member of staff couldn’t do complex calculations as they neither had the information nor the capability. They were dependent on what the finance team would agree to do.
Today, if you want to try to prove something with numbers, it’s taken for granted that you have the tools to do so. You can just open Excel and get on with it.
So it will be with creating software services.
Focus on outcomes
AI will also make us far better at focusing on what matters. At the moment, transport drowns in strategies. We define our outcomes so thoroughly that we no longer know what they are.
I’m working with a transport company at the moment whose CEO is really keen to have one memorable outcome metric. She’s absolutely right: it’s far, far better that the team remember one metric properly than fail to remember a dozen (let alone a hundred).
But our organisations are complex. So it’s even better if the team can remember one metric and the AI keeps track of everything else we’ve actually said we want, so we can filter all our activities through it.
It might even make us better at procurement.
If we could specify solutions actually designed to achieve what we have told ourselves we want, we would get far better outcomes. But that’s inconceivable (too many outcomes!), so instead we specify what we got last time. Nobody writing a specification can keep the whole strategy in their head, so they copy the last spec for ease.
AI can be the answer to this never-before-solved conundrum. It can read everything we have said we want and be the repository for all of our outcomes. We can use this to test our projects, our policies and our specifications.
Of course, it’s all rather weird, as the strategies will themselves have been largely written by AI. That isn’t a prediction: Government strategies in other policy areas are already being written by AI, based on textual analysis by AI detectors.
The future is likely to consist of robots communicating with robots to generate specifications for outcomes read by suppliers’ robots.
If that sounds dystopian, don’t blame me: I’m just the messenger.
But it’s also possible it will result in better outcomes.
Simplifying paperwork
Transport generates a lot of paperwork. Much of it is repetitive and predictable. In a parallel universe we could scrap some of it, but that’s surprisingly hard. After all, just because the transport plan for a new housing development is probably a boilerplate, templated consultancy job with no imagination doesn’t mean that telling the developer that they don’t have to think about transport at all is the right idea.
My podcast guest from a couple of weeks ago, Andrew Browning, has built an AI-enabled business to automate repetitive paperwork.
I can’t be certain that Andrew’s firm is the way this will end up happening (though Andrew’s a seriously bright guy) but it’s certain that AI will generate far more of the paperwork in future, and humans will do far more of the checking.
Contact apocalypse
One of the more immediate consequences of AI is the overwhelming of contact centres as, suddenly, everyone is enabled by their own chatbots to apply for every penny of compensation they are owed and to appeal every fine or penalty they are given.
Way, way back when large language models were new, there was a theory that contact centre staff would be replaced with AIs.
Hmmmm - I doubt it.
I suspect it’s much more likely that contact centre staff will go from replying directly to customers to managing teams of AI agents that are dealing directly with customers. The work will remain but it will fundamentally change in nature.
This is another example of the dystopia of robots talking to robots. Sorry, it’s coming, like it or not.
Decision support
The big difference between AI and every previous technology is that it has judgment.
Excel does sums and a database stores things. Only humans have previously been able to have an opinion and make judgements. As an opinionated bastard myself, I rather resent the competition.
But, hey, if it’s here, we might as well make use of it.
Our organisations are not good at making decisions at the moment. Sometimes we don’t want to disagree with the boss. Sometimes the boss doesn’t want to have an opinion and so keeps asking for ‘more information’ or for it to be signed off by a committee, board or anyone-but-me.
AI doesn’t care who the boss is and can offer up actual decisions to be taken. It won’t be right every time but maybe having a robot that prompts a decision is better than no decision at all.
Training and Personalised Development
Ready for the jobs apocalypse?
I’m not convinced. Technologies can take away work yet, somehow, generate more of it. Washing machines were meant to liberate us from laundry lines and mangles. And they did. Yet we spend just as much time washing clothes because it’s no longer acceptable to stitch yourself into the same underwear in the autumn and only change your pants in the spring.
Excel made financial planning far, far (far!) quicker than the slide rule. So did financial planning end up requiring fewer accountants and modellers? Did it hell! We just did more modelling.
And even when technology genuinely does reduce demand for labour, we just invent new jobs.
My grandad, were he still alive, wouldn’t have a clue what a barista or an influencer was. Yet there are roughly the same number of people working across those two professions as there were coal miners in the early 1980s; a job he would definitely have understood.
So I’m not worried about jobs in total but many, many jobs will change. Luckily, I’m actually quite excited about AI’s role in personalised training and development. I wrote about that in more detail here.
It’s all down to culture
Is AI an opportunity or a threat?
Yes. That’s right. Both.
Big organisations are increasingly swamped with busywork: everyone writing documents for each other and nothing getting done.
Solutions like Andrew Browning’s create the opportunity to take the busywork away and let people focus on outcomes. Rapid prototyping enable customer-focused solutions to be developed much faster. Decision support could stop senior leaders being the bottleneck.
But there’s an alternative reality in which humans simply become slaves to the document-producing machines, which can now produce infinite numbers of documents.
Which of those we get isn’t a question about the technology.
AI amplifies whatever culture you already have. An organisation that’s good at making decisions will get faster. An organisation that’s bad at making decisions will get slower, because it will have more to read, more to check and more to argue about.
If leaders measure activity and process compliance, AI will let them generate more of both than ever before.
If they focus on making a difference to customers, AI will remove huge amounts of friction and free people to do their most meaningful work.
So the question for transport isn’t whether AI is ready for us. It’s whether our organisations are ready for AI.
It’s all down to culture. Like it always is.
