AI at Work: Why EHS Must Be Part of the Conversation

Charlotte Buffoni (00:02.252)

Welcome to Rethinking EHS Global Goals Local Delivery. As workplaces evolve, EHS is increasingly looking beyond traditional physical hazards to consider how work itself affects psychological health and safety. Workload, organisational change, leadership, role clarity and workplace culture can all influence psychosocial risk. And so new technologies, including AI enabled workflow and performance management tools, are adding another dimension to that conversation.

In this bonus episode, we're going to dive a little deeper by looking at one specific example of how the changing nature of work can introduce a new dimension to psychosocial risk, the growing use of AI enabled workflow, reporting and performance management tools. I'm joined by Georgia from Peter Jay Ramsey and Associates in Australia. Thanks for joining me.

One of the interesting things about psychosocial risk is that it isn't always created by an obviously unsafe workplace. It can emerge from changes that are intended to improve performance. Let's take a simple example. An organization introduces AI-enabled workflow reporting or performance management tools. Its aim is to reduce administration, improve productivity, and give managers better information.

Giorgia (00:54.909)

Exactly.

Charlotte Buffoni (01:21.176)

But after implementation, employees begin to report some unintended consequences. Their work feels more closely monitored, productivity expectations increase. Some employees aren't clear about how the data is being used to assess their performance. Managers interpret the information differently and in some areas, the technology actually added reporting requirements rather than reducing workload. So Georgia, looking at that scenario through an EHS lens, what would concern you? At what point?

Does this stop being simply a technology or a change management issue and become a psychosocial risk issue?

Giorgia (01:57.502)

Bye.

Is that the organization has introduced an AI management tool without explaining its role. so an AI risk isn't really just about the technology that it improves, it produces or the business productivity, etc. If you use an AI tool incorrectly, it can definitely negatively impact psychosocial factors of the individual as well as the workplace. because it starts to get into this grey area where it becomes its own.

Individual worker, or kind of like an assistant, they call them agents, well that's what the AIs call themselves, I suppose. but what it does is it reduces worker autonomy, it increases surveillance, it influences decisions, and you if you don't know what you're doing and

Information it produces means and implies and all the risks and flaws from that information, the the outcome can be really detrimental. So if you're relying on an AI to make a business decision, you need to understand that the AI has no idea. The AI is designed to encourage you to use it more. So if you're introducing these kinds of

pseudo supportive tools and not really understanding any kind of safety culture around using that. Th there's entire schools of thought about AI safety, and that's for a good reason.

Giorgia (03:37.374)

But if you're not doing that your own due diligence, there is a really high risk of influencing psychosocial hazards. so for some workplaces like hospitality, AI can be used to manage what do you call them, working schedule, scheduling, rosters, etc. But the AI has no idea that if you roster John Smith for seven hours straight after working seven hours yesterday and the day before on the day before, that he might get really tired and he might, you know, do some accidental dangerous things.

Because the AI says, why can't we optimize this? Like the data says we need to do the absolute maximise. Therefore, this is the correct solution. But the human condition is is way more nuanced than that. and I think acknowledging and understanding and

making an effort to integrate that thought, school of thought into these kinds of AI and digital and any kind of technological kind of influence in the workplace is incredibly important.

Charlotte Buffoni (04:39.533)

So what warning signs should leaders, EHS teams be looking for before these kind of issues could potentially come more serious?

Giorgia (04:49.191)

Yeah, so potential warning signs for AI or technology misuse is a reduced level of role clarity. So if an AI is simplifying the workflow, you kinda kinda be like, well, what part of this is me being a a scientist per se? What part of this is me being a marketing lead? Because the AI can just generate the image for me, it's done. Now what do I do?

And I think what should happen is management, leadership, you know, C suite, everyone higher up should talk to their workers about using AI as a supplementary support but not a replacement for your job. for the auditing sphere, it's impossible to use a robot to audit a business because the robot hasn't seen

you've seen it hasn't understand the nuances and the inter complex the interdimensionality of everything that you're working with. some people don't even know that an AI can't read a PDF. It can't read a Word document, it has to do special Python scripts, it can't get information from a website. It's incapable of doing that. But it will tell you it can. so it's kind of really important that you don't rely on it for anything like legal, anything

You know, it's just it's very new still, it's extremely new and it's it kind of produces this thing called the Jevons paradox where it increases your efficiency but that doesn't mean you're getting more time back. You actually

Do more because now you have more time and that causes fatigue on the workforce. That cause it causes confusion. That causes scrutiny, surveillance, it causes all of these negative detrimental impacts. And all it was is we've just decided, we're just gonna use chat GPT. That's all it was. But it just has such a big trickle down effect. and it's so important that that's understood.

Giorgia (06:55.537)

before any any kind of influence. So yeah, I think that it's really important for all workplaces from the management side down to understand its impact on this workforce.

Charlotte Buffoni (07:08.727)

Yeah, super interesting. I think, you know, thinking of those unintended consequences, you know, I think we all think using AI, you know, can make us much more efficient. And in many ways, yes, it provides us with those tools, but it's still that human element that, you need that interaction and you also need to think of the effect that it has on, yeah, on the workforce. So we've looked at what potentially could go wrong.

when you implement new technology changes and that affect the way people are managed or measured or expected to work. Let's perhaps turn that around and look at what good practice could look like. So if an organization is introducing an AI enabled workflow reporting or performance management tool, what should the organization be considering from an EHS and a psychosocial risk perspective to make sure those human factors aren't missed?

Giorgia (08:01.243)

Yeah, so I think the human factors and systems perspective is always keeping an objective objective approach to AI. As soon as you 'cause you can train them and it it it's just it's really hard to emphasize how much the AI wants you to keep using it. It's going to tell you, I know the answer, this is correct, and you're gonna say, Ha, hang on, no, I know the law.

that is not true and it's gonna be like you got me actually and it's just gonna keep going through that and human brains are designed to see this this authority and be like yeah they must know what they're talking about it's got citations it's got sources so I think from approaching it in a human factor and using it in a positive kind of way I think like in my own auditing I use it to spell check me I use it to

Charlotte Buffoni (08:32.557)

you

Giorgia (08:57.637)

sense check me.

force is doing X and I'm saying have I considered all the factors and it and it will say to me well have you considered this have you considered that and I'm like well I can't consider that because that's not applicable because this is the data center and it's just working with it to brainstorm and it should no way be a supplementary it can't produce reports. It can't produce PowerPoints it can't do that. It's not at the stage of doing it it's incapable of understanding the depth of what you're doing as a human would.

so I think that it's it's so close to being a productive, positive tool. I think it's really the onus is on us to make sure we we're using it correctly.

And for example, in the EHS space, a lot of companies are using it to get the ball ro rolling. So what kind of legislation should I be reviewing? What kind of legislation should I be ignoring? Is this applicable to a major hazard facility? Being able to have that super quick dialogue with the with the chat LLM.

is really helpful. It does increase that kind of thinking. But we're seeing studies where over reliance on these language models are reducing the critical thinking activation of our brain. It's not causing any like deterioration, but it's just not getting as used as it as it typically would.

Giorgia (10:29.457)

So yeah, I think that it's so close to being a productive tool. It's just you have to use it with the critical eye and critical thinking and that's its own skill of itself, so

Charlotte Buffoni (10:40.365)

Yeah, absolutely. And I think you've touched on it there in terms of introducing or managing these kind of changes and it sounds like it's so...

intrinsically linked to that consultation with your employees so they understand the purpose of implementation of these systems, you know, to avoid some of these pitfalls that we've discussed around the unintended consequences. And, you know, I think back, I know both of us working in EHS space, familiar with an ISO management systems perspective, and I think of that consultation and participation requirements. yeah, definitely resonates with me. And I guess

From your perspective, Georgia, what also, yep.

Charlotte Buffoni (11:31.455)

Okay, that's question five on mine, but yeah, that's all right. That's absolutely fine. go to that.

So Georgia, I'd be really interested to hear from your perspective. What role should an EHS professional play in this? Should EHS be involved at the design stage of potential new workplace technologies or rather than being bought in once problems have actually started to emerge?

Giorgia (12:00.593)

Yeah, I think that the integration of and S arm professional in that kind of checkpoint stage. So in in every phase of the development, before an AI or technology is brought onto a project, in the the early stages of that kind of implementation.

The late stages, mid stages, the final product, and then whether or not they're going to continue using it. All of these stages I think are an excellent opportunity for an professional, EHS professional to jump in. And it it will kind of look like alright, so what decisions and what influence will this tool have? Is it going to review a worker's eyes on a monitor?

How are they going feel about that? Is it going to review how many clicks they do? Is it going to change their work roster? Is it going to just be available for use and input? It's really important to do this management of change process, and a lot of businesses I think don't realize that working iteratively through that process is so important because you need to reflect, you need to review how each stage will influence not only the the individual but the

broader business. and then obviously in Australia you have to understand have you even talked to your workers about bringing on this tool? Probably not. It's probably just been a C suite discussion or you know, board directors have said, Yep, done, this do it. and it's just i the other thing is as well is like can you challenge the the use of the tool? Can you say no I will not be monitored. No I do not want my data counter

'Cause if you go too objective and if you think, well, the AI is just helping us do data, it's helping us count KPIs, it's helping us evaluate the employee, then evaluate whether this is a good fit for our business, is having this worker, then you kind of miss like, hang on, no, this staff member gets along really well with the team. This staff member is always contributing to team meetings, all this kind of stuff. And it just it has the ability to remove that person personability.

Giorgia (14:13.44)

so I think that the EHNS professional can encourage whoever's managing managing the introduction of the system or the you know the upkeep of the system to reflect, to step in and to identify what they need to do. It's a hierarchy of controls. You know, do we even need the tool? Has some marketing professionals stood in and said this is the tool for you, or have we actually identified a gap and are filling the gap?

Charlotte Buffoni (14:34.699)

Yeah.

Giorgia (14:42.576)

It's yeah, it's just i I think having that third party perspective is incredibly important to all stages of that introduction.

Charlotte Buffoni (14:49.229)

Yeah, absolutely. So if there's one broader lesson from this example, it's that technology doesn't just change systems and processes. It can change the whole experience and design of work itself. So I'd like to finish by bringing this back to what organizations can be doing now. So Georgia, what's one question every organization should ask when assessing psychosocial risk in their changing workplace?

Giorgia (15:13.338)

I think asking would we be comfortable in explaining to the worker how this system has influenced decision about them?

I think that's the key kind of question. If we say ChatGPT told me that your KPIs are shocking and we're letting you go, that's awful. so I think if you can't clearly explain what the data's for, what the tool's for, what the goals are as an organization, how we've reviewed that, how we've assessed that, how we've come to this conclusion, how we've decided this is the correct way forward, I think if you're not asking those questions

Then what are you doing?

Charlotte Buffoni (15:56.033)

Thank you, Georgia. I've really, really enjoyed the conversation today. Perhaps the key takeaway from today's conversation is that when organizations change the way work is designed, measured or managed, that they also need to consider how that change actually affects the people doing that work. The question isn't simply does the technology work? It's also what has changed for the worker?

Workload, autonomy, expectations, role clarity, leadership and trust, they all matter. And understanding those human impacts is increasingly an important part of effective EHS management. Thank you for listening to Rethinking EHS Global Goals Local Delivery. As Inogen Alliance celebrates 25 years, this series brings together global collaboration and local expertise to explore the changing challenges facing EHS professionals around the world. Until next time, thank you for listening.