As governments around the world race to lead the artificial intelligence (AI) revolution, the focus has largely been on funding, workforce and infrastructure related issues. Yet a growing body of evidence shows that public trust, not just investment, will determine which countries succeed in building sustainable AI ecosystems. For Australia, this means policy must prioritise trust as a core economic and governance objective, not a secondary consideration.

Australia is investing heavily in AI through national strategies, funding programs and institutional initiatives to position the country as a global leader. The National AI Plan recognises that economic opportunity must be balanced with safety, accountability and public confidence. Yet public attitudes toward AI remain mixed. While many Australians recognise its potential benefits, concerns are most acutely centred on workforce displacement, particularly the risks of job loss, difficult transitions, skill mismatches and uneven impacts across sectors and regions, alongside broader issues such as privacy, misuse, fairness and accountability in how AI systems are deployed. These concerns are not unique to Australia. They reflect a broader global pattern in which trust has become a central factor shaping the adoption of new technologies.

Why trust matters for AI outcomes

Our recent study, based on data from 40 countries, provides further insight into this dynamic. We find that AI startup ecosystems are shaped not only by government support, but also by how societies perceive the advantages and disadvantages of science and technology. The key finding is clear: public perception has a measurable impact on AI entrepreneurship. A one-unit increase in negative public attitudes toward science and technology is associated with around a 15 per cent decrease in AI startup activity.

To measure public perception of science and technology, we rely on four questions from the Science and Technology section of the World Values Survey (WVS). The survey data are collected from samples of citizens in each country. For example, one of the relevant questions is: “One of the bad effects of science is that it breaks down people's ideas of right and wrong.” Responses to these questions range from 1 (completely disagree) to 10 (completely agree).

Importantly, not all concerns carry the same weight. Concerns about moral and ethical implications, such as whether the technology undermines social values, have a stronger negative effect on startup formation than concerns about practical risks, such as health impacts. This suggests that trust is not only about safety, but also about alignment with societal values. For policymakers, the implication is clear: AI policy must address both capability and legitimacy. Without public acceptance, even well-funded innovation systems may fail to generate meaningful outcomes.

International examples of building trust

Evidence from our study illustrates this pattern. Countries such as the United Kingdom, Sweden and the Netherlands, where public concerns about science and technology are relatively low, tend to exhibit stronger AI startup activity. In contrast, countries such as Armenia and South Africa, where public concerns about science and technology are relatively high, show weaker entrepreneurial outcomes in the AI sector. This highlights how differences in public sentiment over science and technology generally translate into measurable differences in innovation performance.

The European Union’s AI Act adopts a risk-based approach to regulating AI, with a strong focus on transparency, accountability and protecting citizens’ rights. Similarly, the United Kingdom’s AI Safety Institute focuses on evaluating advanced AI systems and assessing potential risks to support safe and responsible deployment. In both cases, trust is treated as a deliberate policy objective rather than an assumed by-product of innovation.

At a global level, the OECD’s AI Principles similarly emphasise that AI systems should be “trustworthy”, grounded in transparency, accountability, and respect for human rights and democratic values, positioning trust as a core objective of AI policy rather than a secondary outcome.

Policy priorities for building trust in Australia’s AI ecosystem

For Australia, this has immediate policy implications. First, the Federal Government should expand investment in AI literacy and public communication, particularly through the National AI Centre and education systems. This could include integrating AI literacy into school curricula, expanding public awareness campaigns through the National AI Centre, and supporting community-based programs such as workshops, short courses and public seminars that explain how AI systems work and where they are used. For example, targeted programs for small businesses and public sector workers could help build a practical understanding of AI applications and risks. Research shows that when people lack understanding, technologies are more likely to be perceived as opaque or threatening. Improving access to clear, accessible information can reduce uncertainty and build informed trust.

Second, federal regulators should move from high-level principles to enforceable standards for responsible AI. In the next 12–18 months, this could involve introducing mandatory transparency requirements for high-risk AI systems, requiring organisations to disclose how AI-driven decisions are made, and establishing independent oversight or audit mechanisms for sensitive applications such as healthcare, finance and public services. Clear compliance guidelines and enforcement mechanisms would help translate existing principles into practice. Evidence from industry reports suggests that organisations with stronger AI governance frameworks tend to see higher adoption and better performance outcomes.

Third, state and territory governments should use public sector procurement as a lever to embed trust in AI systems. In practice, this could involve updating procurement guidelines to require that vendors demonstrate compliance with ethical AI standards, including explainability, data governance and bias mitigation. For example, government tenders could include mandatory criteria requiring suppliers to provide evidence of risk assessments, transparency measures and human oversight mechanisms before AI systems are approved for use. By requiring that AI technologies used in public services meet strict ethical and transparency standards, governments can shape market behaviour and signal expectations to industry. This could be operationalised through updated procurement frameworks and evaluation criteria that explicitly incorporate trust-related requirements into contract selection and performance monitoring.

Fourth, governments should institutionalise public engagement in AI policy design through mechanisms such as citizen panels or structured consultations, particularly on high-salience issues such as workforce displacement, job transitions and the future of work. Evidence suggests that public concerns, particularly ethical and cultural ones, are not peripheral; they can materially shape whether innovation is accepted, adopted, or resisted. 

More broadly, research shows that people evaluate new technologies based on how well they align with their values and beliefs, rather than on technical features alone. Policies that emphasise fairness, accountability and inclusion are more likely to gain public support and enable broader adoption.

As countries compete in the global AI race, funding, government support, digital infrastructure and AI talent development will remain important, but they are not enough. AI ecosystems are built not just in laboratories and boardrooms, but in the minds of citizens. For Australia, the policy challenge is clear: building AI capability must go hand in hand with building public confidence. Without trust, even the most advanced technologies may struggle to take root. With it, Australia has a far greater chance of translating its AI ambitions into real economic and societal outcomes.

Dr Amir Andargoli is a Senior Lecturer in Information Systems at Swinburne University of Technology and Course Director of the Master of Business Information Systems. His research focuses on artificial intelligence, digital transformation and innovation in organisations, with particular emphasis on how firms develop capabilities to adopt and leverage emerging technologies.

Dr Hassan Gholipour Fereidouni is an Associate Professor of Property in the School of Business at Western Sydney University (WSU) in Australia. Hassan is an active researcher in the fields of property economics, urban studies, tourism economics and Middle East economics.

Professor Reza Tajaddini is the Head of the Department of Accounting, Economics, and Finance at Swinburne University of Technology. His research interests encompass geopolitical economics, behavioural finance and economics.

Acknowledgement
The authors acknowledge the contributions of Mohammad Reza Farzanegan and Fredrick Chege to their article published in Technovation.

Image credit: Canva