Battling bottlenecks: Transport issues AI can (and can't, yet) fix

How are AI and smart technologies transforming the rail and air sectors?

There is significant uptake of AI in the aviation sector—from ticketing and sales to airport management, including gate assignment, passenger flow, and overall passenger experience.

Patrick R. Atack, Transport Editor, GlobalData

Digital twins are starting to make an impact in maintenance, repair and operations (MRO) and construction. When programmed properly, AI systems can help engineers quickly understand and resolve issues in building and infrastructure design.

In maintenance too, the ability to predict when aircraft or trains will need work or checks is likely to boost system reliability. Many tech firms offer this kind of product, but full integration and staff training will likely take 5–10 years.

How is AI helping to improve operational efficiency and reduce delays in rail and air transport?

Increased data collection, interpretation, and management are already improving airport operations through tools like Amadeus and AIRHART, and this will likely accelerate.

This plays out on both the passenger and operational sides. AI-enabled cameras track queues and bottlenecks in real time, while turnaround tasks—cleaning, unloading, refueling—are being scrutinized for efficiency. Why do they take so long, and how can they be improved?

Efficiency can be traced backward from the gate or platform to the routes passengers and staff use to get there. It’s no use streamlining cleaning if staff can’t reach the aircraft due to road closures, or if train cancellations mean the station is blocked. Systems need to be fully communicative—with humans too—and able to understand when timelines can be adjusted because departure isn't possible anyway.

Still, there’s a limit to how much AI can help. It can detect or foresee delays, but it can’t always take action to prevent them. If a vehicle breaks down due to a human or mechanical issue, AI can’t “magic up” another train or plane. It can improve communication and notify systems and humans, but it can’t drive a train along a restricted track to fill in.

There’s also a broader question about AI’s efficiency. It’s often claimed that AI can reduce railway energy use by over 20%, but that figure doesn’t account for how much energy AI itself consumes. Some European cities are already worried that data centers powering these systems may strain public power supplies. If we need huge amounts of energy to run systems that make transport more efficient, are we really gaining—or just shifting the problem?

What are the biggest challenges transport operators face when trying to integrate AI and smart systems into legacy infrastructure?

French train manufacturer Alstom and others have highlighted challenges like a complex regulatory framework, as governments work to decide what systems they do or don’t want—and how to control them. There's also the “black box vs. white box” issue. In transport, visibility into how AI makes decisions is essential, especially where safety is concerned.

This limits how AI can be used in safety-critical operations. As a result, it’s more common in areas like passenger flow management or service reliability.

Public trust is another issue. For every person comfortable with ChatGPT or other tools, there are just as many who don’t want AI involved in their travel—or who don’t trust it with personal data. That comes up in every story I write about AI and transport, particularly around biometrics. For now, the industry is offering opt-outs, but as global trends move forward, there’s a risk that some people will be left behind. We’re already seeing this in sectors like banking and public services, where smartphone access is assumed, and those who can’t—or won’t—use the tech are excluded.

In aviation, are AI-driven tools like dynamic pricing, passenger flow prediction, or smart baggage tracking becoming standard? Is a new default baseline being set for providers and customers alike?

These tools are in use, but they’re far from standard. Companies like FLYR, a travel-tech firm, are working to improve customer experience, but commercial constraints mean they often partner with smaller, luxury-focused airlines that are more willing to adopt new tech.

The process often starts with dynamic pricing and continues through the sales funnel—using apps to upsell luggage, hotels, connections, and more. These firms hope to capture the entire passenger journey with AI-powered sales platforms, but how quickly and widely that happens remains to be seen.

Passenger flow and airport management tech is evolving more rapidly, but often only at one end of a two-ended journey. A passenger might check in their bag with a facial scan and breeze through a smart terminal in a major hub like Tokyo or Frankfurt—but if their destination lacks compatible tech, the process breaks down. Either they don’t get their bag at the other end, or they’re forced to opt out of the system.

The same is true of customs and border control. Countries like China, Canada, and Singapore are exploring biometric-only systems, but the international agreements needed to make those systems work across borders take years, even decades. So, this won’t be ready the next time you or I fly.