Making transport AI ready: why unlocking existing data matters now

Artificial intelligence (AI) is not just reshaping daily life, it is redefining how systems think, learn and respond. Nowhere is that shift more visible than in transport. From predicting disruptions to managing flow minute by minute, AI is already proving it can make networks safer, smoother and more efficient. But it cannot do any of this by itself. It relies on data; accurate, accessible and connected data. And in the UK, that is where both the challenge and the opportunity lie.

Our new report for the Department for Transport, Transport Data for AI Innovation (March 2026), shows the UK is not short of data. Far from it. Across roads, rail, freight and people movement, this country generates an extraordinary volume of transport intelligence every day. The problem is not scarcity; it is fragmentation. If the next era of AI enabled mobility is going to deliver what it promises, the priority is no longer finding more data, but transforming, integrating and mobilising the data we already have and using AI to turn that into information.

In the report, we found that across the 156 datasets we analysed, five recurring “data families” appeared repeatedly across modes, regions and data owners. We also found that if organisations examined all five at the same time, rather than treating isolated datasets separately, the potential impact increased significantly. This parallel focus unlocks larger gains in safety, reliability, customer experience and value for money, and provides a stronger foundation for AI systems that can actively improve network performance.

This raises a sharper question for any transport operator or authority: do such organisations have the tools, capability and agility to turn data into something AI can actually use, not just for the organisation itself, but for the wider, integrated transport system?

Where AI is already delivering

AI is already working inside UK transport systems, but not evenly. In the report, our team found clear examples of places where AI-driven approaches are maturing quickly, especially in larger urban transport authorities and high-volume corridors. Models are forecasting congestion with growing accuracy and detecting stopped vehicles in near real time. Cities are beginning to combine mobile phone data with live traffic signal information to validate conditions on the ground and support minute by minute operational decisions. Some public transport operators are using computer vision analytics on their existing CCTV networks to understand crowding, boarding behaviour and dwell times, enabling interventions where they matter most.

But the pattern is not uniform. Adoption varies significantly by sector, region and organisational capacity. Some authorities and operators are already experimenting with integrated analytics and early digital twin capabilities. Others are still building the basic data foundations needed for AI to be useful. This uneven landscape explains why pockets of innovation exist, yet system-wide benefits remain unrealised.

What holds progress back

If these opportunities are so visible, why is progress uneven? In the report, we identified fragmentation as one of the biggest barriers. Many of the data feeds AI need remain locked in systems that were never designed for external use. Real time signal phasing and detector outputs often sit inside vendor-controlled environments that were built decades ago. High value private data, including last mile logistics information and mobile operating system signals, is hard to access at the scale required even when the public benefit is obvious.

There are also structural issues. Procurement rarely mandates open standards-based data access. Metadata and formats vary widely across authorities, modes and suppliers. Data quality is inconsistent. Many organisations lack the capacity to turn raw feeds into intelligence that can be used operationally. In the report, we found that this fragmentation slows down innovation, raises integration costs and limits what AI can learn. It leads to duplicated effort, longer delivery cycles and missed opportunities to improve safety, reliability and customer outcomes.

AI is already proving it can make networks safer, smoother and more efficient.

Start here: the five data families with the most potential

While this fragmentation may appear daunting, particularly for organisations working out where to begin, our analysis offers a clear starting point. In the report, we found that although data is spread across hundreds of sources, progress does not require tackling everything at once in isolation. Instead, by concentrating on five families of information that are already available in some form, transport bodies can make significant, measurable advances in a focused and practical way:

1. Operational logistics data. Real time courier and haulier operations reveal freight patterns as they happen. When integrated with network management, this helps cities keep people and goods moving during periods of peak demand, disruption or constrained capacity.
2. Mobile operating system sensor signals. High resolution data from everyday devices, including location, motion and inferred mode, can illuminate how and why people move. Used responsibly, this strengthens understanding of demand and supports more targeted interventions.
3. Highway CCTV analytics. Computer vision applied to existing camera estates turns video into structured intelligence on safety, flow and behaviour. This allows targeted action at junctions, crossings and corridors rather than broad measures that may not be effective.
4. Urban Traffic Control and connected infrastructure data. Signal phasing, detector status and configuration information enable dynamic management that responds to live conditions rather than pre-set timings. This is also an essential building block for digital twins.
5. Connected vehicle telemetry. Telematics and safety events help predict conditions, identify risk and support preventative action. These insights also strengthen the evidence base for maintenance, investment and future planning.
   
   

What transport organisations need to do now

Moving from data discovery to data mobilisation is what needs to happen now and there are a number of changes required to enable that. Procurement should require open, standards-based data access from the outset so today’s investments do not become tomorrow’s closed systems. National approaches to metadata, licensing and privacy will help organisations share data once and reuse it many times. Where shared solutions make sense, such as mobile phone data capability with standard interfaces to local traffic systems, they should be developed collectively rather than repeatedly in isolation.

At the same time, organisations need to invest in capability. In the report, we highlighted the importance of building skilled teams able to govern, integrate and exploit data confidently, as well as forming practical partnerships with private data holders that unlock value without compromising commercial realities or public trust.

AI will not transform transport through potential alone. It will do so when the UK’s rich, existing datasets are discoverable, trusted and easy to combine. The foundation is already in place. The task now is execution: unlocking the data, aligning the rules, building the skills and modernising the tools needed to use it.

You can read the full report here, along with the supporting Appendix.