A new tool to ensure AI in workplaces is acceptable and safe

Story written by Dr Sarah Keenihan, AIML

Artificial intelligence (AI) promises a lot when it comes to improving workplace efficiency. Technology may take over mundane tasks in administration, provide better pricing recommendations to sellers or streamline recruitment, as just a few examples. 

But what about the people involved? How will AI affect them? Dr Zygmunt Szpak from the Australian Institute for Machine Learning is developing a tool to find out. 

Working with the South Australian Centre for Economic Studies, the Australian Industrial Transformation Institute at Flinders University, and the NSW Centre for Work Health and Safety, Zygmunt is exploring what sorts of health and safety risks in workers might arise if AI is incorporated into a workplace. 

“We’ve already got systems in place to manage the physical risks of technology – for example, safety around robots in vehicle production lines and automated dispatch centres for large retailers like Amazon,” said Zygmunt.

“What we don’t have yet are processes in place to manage the psychosocial risks of technology.” 

Here, Zygmunt is referring to aspects of work such as disruption in daily routines or reductions in human autonomy. What might it feel like if an algorithm is creating work rosters, and you’ve got a reduced opportunity to have a say in how your weekly hours are scheduled? 

“What we’re trying to do is develop some sort of scorecard where people making decisions about adopting AI in a workplace can do an impact assessment and identify any potential issues that might arise,” said Zygmunt. 

The starting points for this work are ethics principles around the use of AI that various countries and organisations have developed in recent years – including things like accountability, transparency and explainability of the AI. For example, Data 61 has published Artificial Intelligence Australia’s Ethics Framework: A Discussion Paper. 

“Frameworks and principles are a good idea,” said Zygmunt.

“But the problem is that it’s not always clear how these should be translated into practise when we use AI in workplace settings. And so that’s what we’re working on now.” 

Zygmunt and his colleagues are currently conducting detailed interviews with people working in and around AI to get a sense of the scope of issues and disruption that might arise when AI becomes broadly adapted into workplaces.

The insights obtained from the interviews will inform the development of the AI workplace tool for managers and business owners.