AI adoption is only part of the productivity story. So what separates the industries getting the greatest gains?
The tech industry is a natural fit for AI.
It birthed the technology. It employs many of the people building the tools. And it operates in areas where generative AI can be applied powerfully.
So too is media. Its workers deal in words, images, ideas and information – many of the very things generative AI is designed for.
Perhaps more surprising is the industry alongside them at the forefront of Australia’s AI transition: finance, which due to its reliance on data and analysis makes it similarly well suited to AI.
New research from The AI Breakpoint, a national study of 1,006 Australian workers aged 18 to 64, finds employees in financial and insurance services are the most likely to report productivity gains from AI (71%), followed by workers in technology, media and telecommunications (63%) – referred to here as finance and tech and media.
Around three quarters of workers in both tech and media and finance use AI at least weekly, compared with around two in five in government and public administration and retail and consumer services.
A similar pattern emerges among workers who say AI saves them time. More than three quarters of workers in both finance and tech and media say it saves them time, compared with less than half in government and public administration and hospitality and tourism.
Work quality follows a similar trajectory. Two-thirds of workers in tech and media and finance say AI improves the quality of their work, compared to just one in three in government and public administration and retail and consumer services.
So what are finance and tech and media doing differently to spread the benefits of AI more widely among their workers?
Part of the answer lies in the work itself. Both groups report high levels of knowledge-based activities, with finance workers particularly concentrated in administration, research and analysis, compliance and process-driven work, while tech and media workers are more concentrated in technical, digital and coding work.
The nature of work in each industry therefore contributes to how AI’s productivity benefits are distributed. This includes what tasks workers are actually doing with AI to carry out these different types of work.
As the heatmap below shows, AI use among finance and tech and media workers is broad rather than confined to a small number of tasks. Both use AI across research, communication, analysis, writing and other everyday work tasks, although some differences emerge in where each uses it most heavily.
Finance workers are more likely than tech and media workers to use AI for data analysis (44% vs 31%) and meeting notes and transcription (33% vs 20%). Tech and media workers, meanwhile, are twice as likely as finance workers to use AI for coding or technical work (34% vs 16%) and more likely to use it for creative work or design (25% vs 15%).
Interestingly, AI use for data analysis extends well beyond traditionally knowledge-intensive industries. Workers in construction, trades and property (30%) and manufacturing, transport and logistics (31%) report using AI for data analysis at similar rates to tech and media (31%).
A manager in finance aged 35 to 44 in metropolitan Victoria said AI had increased his productivity and efficiency.
“It is becoming a useful tool for handling repetitive or time-consuming tasks like summarising information, drafting communications, and helping with initial analysis of data and documents,” he said.
A manager in tech and media aged 35 to 44 in regional Tasmania described AI’s impact on his work.
“It’s been a huge time saver for smashing through the boring documentation stuff and checking code snippets because projects need to operate smoothly,” he said.
These figures suggest the productivity outcomes of these leading AI industries may partly reflect how broadly their workers are applying AI across different aspects of their jobs.
But broad use is only part of the picture. Another characteristic separating the leading industries is the support surrounding that use.
The answer may lie in the organisation
One of the clearest patterns in The AI Breakpoint is that the industries reporting the strongest AI outcomes are also among those providing the strongest AI organisational support for workers.
In finance, 62% of workers report receiving formal AI training, 67% say their workplace has clear AI rules and policies, and 46% rate their workplace as being well prepared for AI.
Tech and media also performs strongly across these measures – 58% of workers report receiving formal AI training, 69% say their workplace has clear AI rules and policies, and 56% rate their workplace as being well prepared for AI.
Finance and tech and media both sit well ahead of most Australian industries in these organisational AI support measures. Meanwhile across other industries these supports begin to fade to where they are lowest in construction, trades and property and hospitality and tourism.
In all industries workers are more likely to report their workplace having rules for AI use than to be providing the training and workplace preparedness needed to support it. This gap is widest in government and public administration.
The significance of these supports is that training helps workers understand what AI can actually do in their role, clear rules reduce uncertainty about where and how it can be used, and preparedness signals overall perceived capability. And when they are considered together with worker outcomes, the industry comparison becomes even clearer.
For The AI Breakpoint, an AI Support Index was created combining workplace AI training, rules and preparedness. It was then compared with an AI Outcomes Index combining worker confidence, productivity gains and trust with AI.
The relationship across industries is evident. Those reporting stronger organisational support – further to the right on the index, also tend to report stronger worker outcomes – closer to the top of the index. Meanwhile, industries with workers reporting lower levels of workplace training, rules and preparedness generally report weaker confidence, productivity and trust with AI.
Finance and tech and media sit towards the stronger end of both measures.
A mid-level employee aged 35 to 44 in regional Queensland working in government and public administration – which sits at the lower end of both measures – described how AI is approached where she works.
“It hasn’t really been utilised in our workplace and there has been no training provided on how to use it, just that it is available,” she said.
While the chart does not prove that organisational support alone causes better AI outcomes, Breakpoint founder Ian Walker said the pattern points to a strong association between AI outcomes and the workplace environment surrounding AI use.
“Technology, media and telecommunications and financial and insurance services show what mature workplace AI adoption looks like,” Breakpoint founder Ian Walker said.
"These workplaces have moved beyond simply giving workers access to AI and are creating the conditions for workers to use it effectively, rather than leaving them to go it alone."
A hospital or construction company may not be able to adopt AI in exactly the same way as a bank, publisher or tech start-up. But giving workers the support to discover where AI can genuinely improve their work may help those industries find their own path to better outcomes.
As this technology rapidly advances, leaders in these industries need to ask themselves whether they have built the conditions to allow such discovery and the benefits which can follow.
Work with Breakpoint
Want to explore a question like this for your organisation? Breakpoint develops original research that uncovers the behavioural shifts shaping markets, audiences and workplaces – turning the findings into strategic insights, thought leadership and media-ready stories.
Download The AI Breakpoint
The AI Breakpoint explores how Australian workers are adopting AI, how experiences differ across industries, and the workplace support associated with stronger outcomes.
