AI & the future of mobility

AI in a traffic control centre predicts how congestion and disruption spread through the network, and compares what possible interventions would achieve. It combines data from detection loops in the road surface, cameras, navigation data and the weather, and forecasts where queues are likely to form over the coming period. On the railway, it calculates how one delayed train affects the rest of the timetable. The system proposes measures such as closing a lane, lowering the speed limit or rerouting a train. The controller chooses. In future, the centre could make small interventions itself, such as adjusting the speed limit or opening a peak-hour lane, and call in the controller only when something unusual happens.

AI fleet control compares alternative plans for deploying vehicles and crews, working within driving and rest times, maintenance schedules and agreements with staff. It predicts how many passengers will travel on a route or flight, so capacity can be matched to demand. If a train is cancelled or an aircraft has a technical problem, it works out within minutes which alternative affects the fewest passengers, taking crew availability into account. The planner assesses the proposal. The next step is planning across operators: in a major disruption, AI deploys buses, trains, taxis and shared transport from different companies together, as one fleet.

Predictive maintenance picks up wear and approaching faults in measurement data, so a part is replaced before it breaks, alongside the inspections that remain mandatory. Aircraft engines send temperature and vibration data during the flight. Along the track, measuring points check the wheels and bearings of every passing train, and sets of points record how much force they need to change position. AI recognises the patterns in that stream of data that come before a failure. The next step: drones and robots doing their own inspection rounds at night, under trains and around aircraft, and a digital twin of every vehicle, a virtual model that keeps track of its condition and predicts when the next service is due.

Mobility as a service brings journey planning, bookings and payments from different operators together in one app. AI weighs more than travel time: personal preferences, live delays and the availability of shared transport count too. If a connection is missed, it recalculates the whole trip, including the onward transport. If the planner is linked to a robotaxi service, which already runs without drivers in a number of cities, it can book that ride as well. For transport operators this opens a sales channel in which their service is part of a combined trip, sold per ride or by subscription. And where data on the whole trip is available, patterns appear that their own ticket sales do not show: which connections passengers choose and where they switch to another mode of transport.

An AI agent carries out a series of steps on its own to complete a task, within limits set by the organisation. In a disruption, thousands of passengers need to know their onward options at the same moment. An agent can find a new connection, arrange a hotel or a replacement bus, keep the passenger personally informed and handle a compensation claim. Automatic rebooking is already in use; airlines are testing agents that change a booking themselves in a conversation with the passenger. The next step is agents that see a disruption coming and rebook passengers before the train is cancelled or the flight is scrapped. That leaves staff more time for the passenger in a wheelchair or the child travelling alone.

Robotaxis already run as a paid service in several cities, with hundreds of thousands of rides a week. Their use remains tied to a defined service area and to the conditions the system was developed for. Within those limits, AI does the driving; a control centre helps vehicles that get stuck in an unusual situation, such as roadworks. The operation covers more than automated driving: helping passengers, charging, cleaning and spreading the vehicles across the service area are all part of it. Data from completed rides yields new training situations, which developers use to improve and test the next software version. With every ride, the collection of traffic situations the system is tested against grows.

In a software-defined car, software determines much of what the car can do, from driver assistance and battery management to the screen in the dashboard. That software is updated remotely. A manufacturer can fix a fault or add a feature without the car going to the garage. More and more cars have a voice assistant built on a language model, which plans the route, finds a charging point or explains what a warning light means. The next step is an assistant that knows the car and acts on its own: it books the service, charges when electricity is cheaper and fits trips around the owner's diary. For car companies, the relationship with the customer carries on long after the sale.

In an AI-first organisation, work on planning, deployment and maintenance starts with AI: it does the work people do not need to do. Staff concentrate on work where experience, judgement and contact make the difference. AI drafts the first version of the timetable, predicts how busy services will be and flags faults. People help passengers and assess situations that call for experience, such as a fault that is not in the manual. The organisation of 2040 could direct a mixed crew: its own people, machines from partners and agents that talk to customers. Along the way, work can shift from routine tasks to direction and oversight.

When vehicles drive themselves and travellers let an agent arrange their trip, jobs appear that barely exist today. In a scenario for the next five to ten years, a fleet supervisor in a control centre watches dozens of self-driving vehicles at once and steps in when something unusual happens. A trainer teaches AI systems to handle rare traffic situations, from a fallen tree to a police officer directing traffic with hand signals. A journey designer builds services that travellers' agents can read and book. Controllers for low-altitude airspace keep drones and air taxis above the city apart. And the planner reviews proposals from a system instead of solving the puzzle by hand.

In tomorrow's traffic control centre, controllers for road, rail and water work with AI as one team. Looking across the networks, AI estimates how a blockage on the railway could push extra traffic onto the motorway, and how an open bridge could make that queue longer. The controllers agree on what to do: run extra trains, keep a lane clear for the emergency services, keep the bridge closed for a quarter of an hour. Each still runs their own network; AI gives them a shared picture of road, rail and water. In an accident or emergency, people take the decisions. Networks that are still watched separately today start looking ahead together.

Transport leaders look a long way ahead. They take the initiative on AI and new technology, work from a picture of the passenger fifteen years from now and show their people what it means in practice. They try things out on one bus route or in one depot, learn from what goes wrong there and make decisions while not everything is settled. That is how they build tomorrow's transport together with their teams. In 2040 they could be running a network in which vehicles, data and agents from other organisations also play a part.

A future academy is an in-house school for the people who already work at the organisation and for those who have yet to join. The organisation prepares them for the future there, rather than waiting for the education system to catch up. Staff learn to work with AI, agents and other new technology, and see what it means for their own tasks. Leaders learn to make choices in an organisation that keeps changing. Because the academy is part of the work itself, the programme changes along with the technology.

An ecosystem means that organisations which each provide part of a service work together as if they were one system. The customer sees one service; behind it are partners, start-ups, scientists, and the public bodies that manage roads and railways and set the rules. Technology comes from one continent, components from another, and the service itself involves organisations in several countries. Data from partners helps AI plan and adjust shared operations. That is why organisations around the world are working on common data formats, so their systems can exchange data more easily.