The conversation about AI in our profession is everywhere. Serious understanding of what it will actually do to the profession is much harder to find.
Ask any room full of HSSE professionals whether they have heard of AI and every hand goes up. Ask them what it will mean for how they work, what they are valued for, and what their role looks like in five years, and the conversation gets much quieter.
I have been watching technology transform professional practice for most of my career. I say that with some authority, having lived through nearly every phase of it.
I have seen this before. Just never at this speed.
In high school, I got my first calculator. It replaced the slide rule, and at the time that felt significant. In university, we learned to program mainframe computers using punch cards, feeding stacks of cards through a reader and waiting to see what came back. During my doctorate in the mid-1980s, I bought my first personal computer: a Turbo XT running on DOS, with a floppy disk drive. When I later upgraded to a 20MB hard drive, I genuinely believed that would be sufficient storage for a lifetime.
My own doctoral work focused on computer modelling, while elsewhere in the university colleagues were developing expert systems: rule-based programs designed to replicate the decision-making of human specialists. We did not call it AI at the time, but that is exactly what it was: an early, clumsy, and genuinely ambitious attempt to encode professional judgment into a machine.
It is easy to laugh at that now. It is less easy to acknowledge that we are probably making equivalent misjudgements about AI today by assuming that what we can see in front of us represents the scale of what is coming.
Over the four decades since that first PC, I have watched the technology landscape of professional practice go through several distinct phases. Simple databases gave way to complex business applications. Those gave way to hyper-specific software tools for every conceivable function. Then came big data, cloud ecosystems, and the integration of systems at a scale that would have been science fiction in 1985. And now, generative AI: tools that do not simply store, retrieve, or process information, but reason across it, synthesise it, and generate outputs that previously required human expertise to produce.
The only constant across my career has been change. And the pace of that change has not been linear. It has been accelerating in ways that remain difficult to fully comprehend, even for those of us who have been watching closely.
Which brings me to the point. The HSSE professionals entering the field today are not facing another software upgrade. They are facing a shift in what professional capability itself means. Most of the profession, understandably given the pace, has not yet absorbed what that actually implies.
The floor is rising, not the ceiling
Most HSSE professionals I speak with have tried an AI tool. Some have asked ChatGPT to help draft a procedure, summarise a regulation, or structure an incident report. The typical response is: “It is impressive, but it made some errors. I had to check everything.” They conclude that AI is a useful but unreliable assistant, good for a first draft and not much more.
That conclusion is not wrong. It is just dangerously incomplete.
When we focus only on AI’s errors and limitations, we miss the structural point. The question is not whether AI is perfect. The question is: compared to what?
An HSE professional in a large multinational operating across the Gulf, managing contractor compliance for five simultaneous projects, tracking legal register updates across three jurisdictions, preparing for two external audits, and handling routine incident documentation, is already operating at the edge of what any individual can competently manage. AI does not need to be perfect to be transformative in that context. It needs to be fast, consistent, and available at midnight on a Thursday when a critical permit is being questioned on site.
The real disruption is not that AI will suddenly make the best HSSE professionals dramatically better. It is that AI is raising the baseline capability of everyone else. That changes how competence is developed, how work is reviewed, and how professional value is measured.
But a rising floor creates its own risks. If AI allows more people to produce work that looks technically competent, the profession needs to become much better at asking whether that work is actually complete, applicable, defensible, and safe to rely on.
Those implications are practical, not theoretical. Organisations will need to rethink how they staff HSSE functions, how they train junior professionals, how they validate AI-assisted work, and what genuinely experienced practitioners must offer to justify their value.
The competency gap is being compressed
For thirty years, a significant part of HSSE expertise has been knowledge-based. Knowing the regulatory requirements for hazardous waste disposal in the UAE. Knowing what a credible corrective action plan looks like after a citation. Knowing the structure of a bow-tie risk analysis. Knowing which clauses of ISO 45001 apply to contractor management.
That knowledge took years to accumulate. It was a legitimate barrier to entry and a legitimate source of professional value.
AI has not eliminated that knowledge. But it has dramatically compressed the time it takes for a less-experienced professional, or a non-specialist who knows how to prompt well, to access it. A competent user of current AI tools can produce a first-draft legal register, a gap analysis against a management system standard, or a structured incident investigation in a fraction of the time it would have taken two years ago.
This is not a future scenario. It is happening now, on real projects, inside the organisations many of us work for or advise.
For HSSE leaders, this creates a very practical challenge. If knowledge access becomes easier, then the differentiator is no longer simply knowing where the answer is. It is knowing whether the answer is complete, whether the source is reliable, whether the context has been understood, and whether the recommended action will actually control risk. The value shifts from information possession to judgment, verification, and accountability.
The “you don’t know what you don’t know” problem
There is another risk that deserves more attention: the “you don’t know what you don’t know” problem. AI can produce outputs that look structured, confident, and professionally credible, even when the user lacks enough domain knowledge to recognise what is missing, misinterpreted, or dangerously oversimplified.
For experienced HSSE professionals, this is a manageable risk because they can interrogate the output. They know which assumptions to challenge, which regulations to verify, which site conditions matter, and where a neat answer may conceal a weak analysis. For less-experienced practitioners, or for managers using AI outside their competence, the danger is different. They may not know enough to see the gaps. That is where AI becomes most risky: not when it is obviously wrong, but when it is plausible enough to be trusted.
This is why AI literacy in HSSE cannot simply mean learning how to write better prompts. It must include knowing when not to rely on the output, when to escalate to a competent person, and how to validate AI-assisted work against real-world risk, legal obligations, and operational context.
Why this is the beginning, not the end
What makes this moment genuinely important, and genuinely unsettling for those paying attention, is that the tools available today are primitive compared to what is coming.
Current AI tools are generalist. They are not trained on your organisation’s incident history, your specific regulatory jurisdiction, your site’s risk profile, or your management system’s particular configuration. They hallucinate. They miss context. They cannot walk a construction site, interview a worker, or read the body language in a toolbox talk.
But the trajectory is clear. The tools are becoming more specialised, more integrated into the platforms HSSE professionals already use, and more capable of reasoning across complex, multi-variable situations. Many of the errors that make current AI easy to dismiss are being addressed systematically, with substantial resources, by companies whose entire business model depends on solving them.
The HSSE professionals who are experimenting now, building practical literacy with these tools, identifying where they fail, and learning how to deploy them responsibly, will be positioned to lead that next wave. Those who are waiting for AI to become reliable before engaging with it are setting themselves up to engage with it too late.
What this series is about
This is the first post in a series exploring what the AI transition actually means for HSSE practice, not in abstract terms, but in operational ones.
We will start with the question most professionals are not asking: when AI-assisted HSSE practice produces a harmful outcome, and at some point it will, who is accountable? From there we will look at where AI fails HSSE professionals and why those failures matter, what it is already doing inside real compliance functions, how the value proposition of experienced practitioners is shifting, and what skills will define the profession’s most effective people over the next decade.
I am writing this from inside the transition, not as an observer of it. Our own work at Redlog has made one thing clear: AI can improve the speed of compliance analysis, but it does not remove the need for expert interpretation, source validation, and accountability. As we expand our compliance tools to incorporate AI that draws on our own regulatory databases, we are seeing where AI genuinely adds value in HSSE practice, where it needs expert oversight to be trusted, and what the liability implications are when it gets things wrong. Those questions shape everything in this series.
The HSSE profession has always adapted to change: to new management system frameworks, to digitalisation, and to the expansion of scope from safety into health, environment, and security. This transition may be larger than any of those. It deserves the same honest, rigorous engagement we bring to any major risk.
The professionals who approach it that way will not just survive the shift. They will define what comes next.
