AI-driven cognitive overload emerges as psychosocial risk blind spot

Artificial intelligence is creating a new category of psychosocial harm in Australian workplaces, as organisations cash in the efficiency dividend of AI transformation before the technology has properly bedded down, according to a leading WHS lawyer.

This is leaving workers to absorb an unsustainable concentration of cognitively demanding tasks with fewer colleagues to share the load, said Hamilton Locke partner, Michael Tooma, who will present at the upcoming AIHS National Health and Safety Conference. 

Tooma suggests that what is commonly called “brain fry” is more accurately understood as cognitive overload, and that AI deployment is accelerating it in ways that current regulatory frameworks are not designed to address.

“What AI is doing in some respects is it is removing a lot of the less cognitively demanding tasks and then concentrating the more demanding tasks into a working day,” Tooma said. 

“But the real issue is that leaderships of organisations are cashing in the efficiency dividend of a wholesale reform of that organisation in the form of structural reform and headcount reduction before the transformation has taken effect.”

The result is that workers further down the hierarchy are experiencing something qualitatively different from what they signed up for.

“The mantra of doing more with less sounds like a really good thing, but typically the way that that is experienced by a person further down the hierarchy in the organisation than the boardroom or the senior executive team is a lot more intensive, a lot more cognitively engaging and tiring work with no breaks,” said Tooma.

The cognitive overload problem is compounded by a second, distinct psychosocial hazard: anxiety about job security. Tooma said workers across industries are openly discussing the prospect of AI replacing their roles, and that psychological harm flows regardless of whether their employer actually intends to cut jobs. 

“It doesn’t even matter if that organisation wants to replace that person or not. It’s the fact that they believe it, and they genuinely feel that anxiety that creates a psychosocial hazard for them,” he said.

He pointed to a Boston Consulting Group study published in the Harvard Business Review, which found that organisations were measuring and rewarding token consumption as a proxy for performance.

“It’s like saying your performance is now measured by how many emails you write. You would laugh at that as a KPI, but yet we are now setting those kinds of KPIs on people in organisations that are not necessarily tech-based.”

The AI dimension, however, is part of a wider argument Tooma will put to conference delegates: that despite a surge in legislative activity over the past five years, and significant penalties handed down to organisations that have failed to act, Australian workplaces are still not looking at psychosocial hazards the right way.

While regulations now require organisations across Australia to identify and control psychosocial hazards, Tooma’s contention is that the tools being applied remain poorly matched to the problem.

In the physical domain, for example, a chemical risk, a crushing risk and an electrical risk are each addressed on their own terms, even though all involve potential physical injury. In the psychological domain, organisations have done the opposite, grouping hazards together simply because they affect the same body part.

“We’ve grouped together a bunch of hazards that are disparate in their character, but have only one thing in common, which is the body part in which they affect, being the brain or central nervous system,” Tooma said. 

“And the solution that we’re applying is a solution that we say is tried and proven: identify the hazard, assess the risk arising from those hazards and implement controls. But in many respects, that tool may not be fit for purpose.”

To illustrate, Tooma pointed to a case involving The Good Guys, in which a worker who sustained a neck injury was subsequently ostracised by colleagues through small acts of incivility that escalated over time, ultimately contributing to her death by suicide. The case demonstrates how conventional risk assessment can miss the early signals that matter most in a psychosocial context. 

“The risk, when it presents as a bit of incivility and a bit of banter in relation to a physical injury, might not present as a high risk from a consequences perspective in an organisation. That in itself might mean that it’s missed or not prioritised in the right way,” he said. “But dealing with the small things in a psychosocial context might actually prevent the big things.”

Another common shortfall for organisations is falling back on superficial wellbeing initiatives, such as gym memberships, fruit boxes or discounted yoga classes, rather than addressing systemic organisational issues. “You don’t earn the right to have a conversation about my personal resilience until you have fixed the fundamental systemic issues in the organisation,” he pointed out.

Rather than focusing solely on risk prevention, Tooma said organisations should invest in building psychological reserves so that workers are better placed to absorb difficult periods. The two foundations he nominates are civility and care: ensuring workers feel valued and respected rather than reduced to an output measure. 

“If you feel that the organisation cares for you, your ability to speak up when you feel that you need a day off for wellbeing reasons is vastly increased,” he said.

He highlighted the importance of job design that builds autonomy, and supervisory behaviour that does not trade off psychological safety against operational output. 

Similarly, Tooma said the problem with AI transformation is not the technology itself, but the organisational design choices surrounding it. Workers need to understand their role in the transition and have meaningful input into what is achievable. 

“We need to stop thinking of AI as replacing people, and that’s the efficiency gain. We need to build up their capacity to use and deploy that tool, and then autonomy in how they use and deploy that tool in a meaningful way,” he said.
On the adequacy of existing laws, Tooma concluded: “It may well be that we need to have a holistic look at the underpinning regulatory framework that deals with these issues rather than simply expect that traditional tools are fit for purpose. The laws are not written in a way that anticipates these types of problems.”

Tooma will present at the AIHS National Health and Safety Conference, which will be held at Adelaide Convention Centre from 15-17 June 2026.