Australia lags US on trust in AI systems, Fujitsu finds

Australian business and technology leaders are less confident in the AI systems used inside their organisations than their peers in the US, UK and Japan, according to Fujitsu research. The survey found 70% of Australian leaders were not fully confident in their organisations' AI.

The findings point to a clear gap between trust in AI and reported business results. In the US, 68% of enterprises said AI had improved productivity and internal efficiency, while 70% said it had improved their ability to develop new products and services. In Australia, those figures were 46% and 50%.

Fujitsu surveyed 400 senior business and technology leaders across Australia, the US, the UK and Japan, all from enterprises with 1,000 or more employees. Australia recorded the weakest confidence levels among the four markets. In the US, 53% of leaders were not fully confident in the AI systems their organisations rely on, compared with 61% in both the UK and Japan.

The data suggests trust in AI is becoming a broader management issue as companies move the technology from small-scale pilots into core operations such as finance, customer functions, software, human resources and supply chains. For boards and senior executives, concerns about whether AI systems can be relied on are tied to risk, investment decisions and the speed at which gains can be achieved.

Data concerns

Poor data quality emerged as the main barrier to confidence in Australia. More than half of respondents, 55%, said it was the single biggest reason they did not trust the AI they had implemented.

That compared with 33% in the US, 35% in the UK and 46% in Japan. The figures suggest Australian organisations face a more acute challenge in ensuring the data feeding AI systems is accurate, structured and usable.

"Australia's ability to compete globally depends on how well we trust and use powerful AI systems. The Fujitsu AI Confidence Research shows that while Australian businesses have the ambition, an AI confidence gap is holding them back. You can't innovate effectively if you don't trust the tools you're deploying," said Mat Franklin, managing partner Oceania - Uvance Wayfinders, Fujitsu.

Australia also ranked last among the four countries on data readiness and AI governance practices overall, even as many organisations have started investing in formal structures to address those weaknesses. The research draws a distinction between recognising a problem and solving it, indicating many Australian businesses are still strengthening the basics of AI deployment.

Governance push

At the same time, Australian organisations were ahead of international peers on some of the measures being introduced to close the gap. Some 62% of Australian leaders said their organisations had advanced strategies for data readiness, ahead of Japan on 56%, the UK on 51% and the US on 43%.

Australia also led in the establishment of formal AI governance and ethics frameworks. The survey found 61% of Australian respondents reported advanced arrangements in that area, compared with 57% in Japan and 48% in both the UK and the US.

Those figures suggest that while confidence remains low, many larger Australian companies are moving to put in place the internal controls, policies and data processes needed to make AI systems more dependable. The gap between current trust levels and governance efforts also indicates organisations may need time before those changes are reflected in day-to-day business results.

Peter Carr, technology analyst and advisor at Councilio, said stronger oversight would be necessary if AI systems are to be used in sensitive or business-critical settings.

"A policy on paper is only the first step. Boards need to establish independent governance systems and strict decision-making guardrails. If we cannot clearly audit how an AI system arrives at an output, we should not allow it to run core business operations," Carr said.

The survey focused on large enterprises, which often have more resources than smaller businesses to introduce governance frameworks and data programs. Even so, the findings indicate that scale alone has not resolved confidence issues and that concerns over data quality remain central even in organisations with significant technology budgets and executive oversight.

The contrast with the US is particularly striking because American respondents were both more confident in their AI systems and more likely to report commercial benefits. That comparison reinforces the idea that trust is not only a technical issue, but also one linked to how quickly organisations can turn AI use into measurable gains in productivity and product development.

Franklin said the main task for Australian organisations was to focus on foundations rather than rush further deployment.

"You can't build trust in AI on a weak data foundation. Investing in data quality and establishing robust AI governance frameworks are prerequisites for any organisation that wants to fully leverage leading global AI models," Franklin said.

He added: "The good news is the problem of low AI confidence can be solved. By focusing on data quality and governance, Australian organisations can build the confidence to catch up and even lead on the global stage. This is how we turn an AI productivity crisis into a national opportunity."