RSSB’s Futures Lab: what generative AI could mean for rail
AI technologies aren’t new, but the recent pace of development—coupled with the increasing level of public awareness—has been significant. GenAI, in particular, has risen to prominence this year. It uses software called generative pre-trained transformers (GPTs) to apply machine learning and deep learning techniques to large language models containing large volumes of both public and proprietary data. Using direct user prompts, GPTs synthesise the data to generate text, imagery, music, and other outputs.
Extensive press and media coverage has shone a light on its advantages and limitations, and its adoption is growing.
The recent investigation into GenAI by RSSB’s Futures Lab has culminated in the release of its first Rapid Technology Briefing. This report, which you can download here, compiles what we know so far to set out a helpful ‘present-state’ description of these technologies. It also highlights key areas for organisations to consider and outlines how the technology may continue to develop.
Caution necessary during GenAI’s continued development
It’s clear, then, that the powerful abilities of GenAI hold promise in lots of organisational areas. With use cases as diverse as content creation, ideation, and threat detection, this technology no doubt has the potential to introduce significant change to the industry. For example, it could:
- streamline the generation of new materials
- reliably assist customers or passengers with their enquiries
- identify breaches in an organisation’s security system.
But there’s another side to GenAI—one that could, if not considered carefully, both exacerbate known risks and introduce new risks to existing systems and processes. Specifically, according to the Rapid Technology Briefing, ‘privacy, data ownership and usage, model outputs, and third-party service dependencies are generating the most prominent concerns’. For example, one key concern around the adoption of GenAI is the challenges associated with inadequate control and storage of organisational data. Without sufficient data security measures, there may be an inadvertent transfer of proprietary organisational or commercial information to third parties or an increased risk of cybersecurity hacking.
All that said, several other sectors are already actively exploring how to utilise GenAI for meaningful purposes.
GenAI in other industries
NASA has already started to use this technology to design parts for its spacecraft. To do this, a worker enters a prompt that informs the system of the geometric data and physical specifications required for the design. The GenAI tool then processes these inputs, compresses everything, creates the design, and, if needed, performs corrective actions. The result is viable, expertly crafted parts.
Aviation has also developed a number of use cases to make the most of GenAI, including:
- examining aeroplane sensor data and maintenance records, which could help predict possible issues and optimise repair schedules
- creating personalised training scenarios for pilots-to-be and offer bespoke feedback on their performance
- enhancing flight routes and improving fuel efficiency by examining prior data and weather patterns
- creating custom travel recommendations for passengers based on trips they’ve taken in the past.
And what’s particularly exciting is that many of these use cases are relevant and translatable to the rail industry. Not only could they support rail’s sustainability goals, but—based on the use cases above—they could also improve driver training, passenger experience, and rolling stock development.
Moving rail forward
Leveraging this technology could revolutionise parts of the rail industry. Key to this will be several activities that rail organisations should consider:
- maintaining a strong level of situational awareness to track how these technologies further develop
- starting a process to identify where they can be put to most beneficial use
- considering how to mitigate the dynamics risks and concerns that currently accompany GenAI.
Based on the findings outlined in the Rapid Technology Briefing, it’s clear that more work is needed to fully understand the evolving and longer-term potential of GenAI technologies. To support this, RSSB’s Futures Lab plans to develop a second and updated edition of the GenAI Rapid Technology Briefing in spring 2024.