At Irwin and Colton’s recent SafetyTech and AI event, hosted in partnership with Make UK and Protex AI, we explored the current state of AI adoption in EHS, where the barriers are and how they can be overcome and close the gap between detection and prevention.
Artificial intelligence was not positioned as a new concept or experimental add‑on; instead, it was discussed as something already rooted within many organisations’ safety systems, with the emphasis placed less on whether AI belongs in health and safety and more on how effectively it is being applied in practice.
With senior leaders from across the HSE community in the room, discussions focused on the maturity and behavioural impact of AI. It quickly became clear that most organisations already have access to AI tools and extensive safety data, and that the real differentiator is not the technology itself, but how it is used, trusted and applied to drive change and give time back to safety professionals.
Chris Newson: The current state of AI adoption in EHS
Chris Newson, of Make UK, opened the day by setting the wider context around how AI is starting to reshape the role of safety professionals.
He noted that while the UK performs relatively well in global safety terms, the opportunity now is less about generating more data and more about how effectively it is used to support better decision-making.
Chris highlighted how much of a safety professional’s time is still absorbed by routine, administrative activity such as updating risk assessments, reporting, and manual data handling. His point was that AI has real potential here, not to replace judgment, but to automate and streamline these tasks so time is given back to the function.
That shift, he argued, creates space for safety teams to move away from transactional work and focus more on driving meaningful cultural change within organisations, where the real long-term impact is made.
Robert McCarthy: Overcoming barriers to AI adoption
Building on Chris’s session, Robert McCarthy from Protex AI explored why AI adoption often stalls even when results are proven. Protex AI case studies showed clear reductions in unsafe behaviours alongside increases in productivity.
The message was simple: Technology only delivers value when organisations act on what it reveals.
Barriers exist at multiple levels. For leadership, the challenge is prioritisation. Even where ROI is clear, AI competes with other day-to-day demands, and without a strong link to risk reduction and business goals, initiatives can remain stuck in pilot mode.
For the workforce, trust is the key issue. Concerns around surveillance, privacy and intent can quickly undermine adoption if AI is seen as monitoring rather than supporting people. Clear governance and non‑punitive use are essential.
There is also a capability gap. Many teams have access to data but struggle to translate its insight into actions.
With all of this in mind, the key takeaway was: AI rarely fails because of technology. It fails when leadership alignment, workforce trust and data usability are not addressed together.
Kieran Bowker: Closing the gap between detection and prevention
After the break, Kieran Bowker, from Pixaera, shifted the focus toward training and competence, particularly in contractor‑heavy environments. Detection, he argued, is working well, but prevention is not keeping pace.
Data can identify unsafe behaviour, but it rarely explains why decisions were made in the first place. Traditional training methods create awareness, but awareness does not translate into muscle memory. With less site experience and limited time away from operational roles, the gap is widening.
Pixaera’s decision‑led, gamified training approach was presented as a way to replicate real‑world judgement under pressure, helping people practise decisions - not just recall rules.
Christian Robinson: AI Assistants in Practice
The afternoon session brought the theory into application. Christian Robinson, also of Protex AI, led an interactive workshop exploring how AI assistants can be used responsibly to reduce administrative burden, improve consistency and support better decision‑making. Importantly… without replacing human expertise!
Christians' session reinforced a central theme from earlier in the day:
AI is most effective when it removes friction from processes, allowing safety professionals to focus on influence, engagement and prevention.
Final reflections
By the close of the event, one message was clear…
Most organisations are no longer constrained by a lack of visibility. They are constrained by their ability to convert insight into consistent behavioural change.
AI is not the solution in itself. The outcome depends on trust and the willingness to act on what the data already shows. Closing the gap between detection and prevention is where the next phase of progress in HSE will be defined.
A quick thank you
Thank you to everyone who joined us at the SafetyTech and AI event. It was great to meet so many HSE professionals interested in technology in the Safety world.
We hope you found the sessions insightful and that the workshop and networking opportunities generated ideas you can take back into your organisation.
Finally, a big thank you to all our speakers and everyone who contributed questions and perspectives throughout the event.