Paul Leach
Head of Human Factors, RSSB
In the rail industry, artificial intelligence (AI) is still in its infancy but is already being applied in areas as diverse as data-crunching, trespassing, and train operations, where it has had a significant impact. AI-powered solutions can improve safety, cut costs, and give rail customers a better experience. We need to ensure that we get the most from AI — and avoid the pitfalls.
AI is not simply a systems issue; it needs to be considered in the broad context of the whole operation. How it is brought in and applied requires careful planning and consideration. AI’s potential can only be unleashed if it can be trusted and governed.
And AI does not exist in isolation. It needs to work for, and alongside, people. The introduction of AI needs to focus not just on functionality but what the human needs it to do. And we need to optimise how an individual interacts with and uses AI.
This is why is it is important to look at human factors guidance principles, that is designing systems and equipment to match human capabilities and limitations.
RSSB’s research team have created the Enabling AI Programme to ensure that the change due to AI is better understood, and standards, safety, and in particular, these human factors are at the heart of enabling the AI revolution.
So, it is worth starting by looking at what people do not do well and what AI could do for them.
This may include repetitive tasks or sedentary vigilant tasks that require close attention for long periods of time.
AI team-mates
AI is also able to process very large amounts of data very quickly and draw connections from a great many factors, tasks that are beyond the capabilities of humans.
And with AI taking away the burden of this work, this allows humans to do the things they are good at. This includes decision-making, being flexible, taking action, spotting risks and hazards, being resilient, seeing when situations aren’t going the right way, and keeping a clear understanding of what’s going on.
One scenario would be the use of digital assistants — not robots but more like AI team-mates. Here, they would quickly calculate all the different things that need to be considered in different situations, and give a human options to choose from, and a confidence rating to enable decisions to be made.
AI will certainly develop a lot more. At the moment, it's still very much in its infancy in terms of what it can do for us. That's one of the reasons we did the work on human factors principles in AI; getting them embedded at the beginning enhances system design and ensures the AI system optimises human performance.
Rail can learn from other safety-critical industries' use of AI, such as nuclear or aviation. And there are areas where rail needs to understand what it specifically needs. But wherever we take our learning about AI and human factors from, ensuring the use of AI includes human factors will improve rail safety and performance and reduce costs in ways that are workable for the human beings delivering our rail service.