AI is transforming our world 

Artificial intelligence (AI) has the potential to unlock transformative economic, social and environmental opportunities for Australia. The potential for public benefit is significant, provided the development, adoption and use of AI is governed in a safe, responsible and sustainable manner. Governing AI in this way underpins community trust and stakeholder support and works to retain a social license. Importantly, good governance of AI also increases the likelihood that organisations will implement and scale up AI effectively and successfully. In other words, good governance creates a virtuous cycle whereby support for the widespread investment in and adoption of AI is maintained, and the transformative benefits of AI are more likely to be realised both at a business and societal level. As Australia seeks to keep pace with international competitors and leaders in the adoption and use of AI, it must also seek to be a global leader in responsible AI.

In Australia and other Western economies, the adoption of ethical AI principles has been an important feature of building robust AI governance practices. In 2019, the Australian Federal Government announced its voluntary Artificial Intelligence (AI) Ethics Framework to guide businesses and governments in designing, developing and implementing AI responsibly. A number of businesses tested these principles in a pilot project overseen by the Department of Industry, Innovation and Science (now the Department of Industry, Science and Resources). The pilot aimed to collate practical learnings and understand the challenges of implementing the principles, which included: 

• Ensuring procured services and solutions are ethically responsible;
• The need to raise awareness of AI ethics, educate staff on the benefits of implementing AI ethics and improve training; and 
• How to judge, measure and continually reassess principles that involve value judgements like ‘fairness’.

The release of these principles in Australia coincided with the peak hype in public discussion, development and adoption of AI ethical principles in industrialised Western economies in early 2019. However, nearly two years on from ‘peak principles’, many commentators were still lamenting the evidence of poor AI governance. This was a common theme in CEDA’s 2020 PubIic Interest Technology Forum.

Prompted by these observations, CEDA undertook a series of roundtable workshops with member organisations in late 2021 and early 2022 to gain insights on the progress towards and challenges of using responsible AI principles and practices within organisations. 

This report summarises key insights from these workshops and is designed to highlight the practical progress and challenges of responsible AI implementation. It also seeks to identify where further support is needed to accelerate implementation.

From our roundtable discussions, it became clear that the level of AI maturity within Australian organisations remains low or evolving. Key challenges highlighted through the workshops included: 

• ‘Siloed’ approaches to AI within organisations; 
• The misunderstanding of responsible AI within organisations, including at the leadership level;
• That governance and ethical use of AI are largely deemed the province of tech teams;
• The difficulty of building an organisation-wide shared understanding and approach to aspects of responsible AI such as fairness and explainability;
• The lack of formal governance frameworks or frameworks that are divorced from broader risk and governance approaches; and
• How to effectively ensure that procured AI services and solutions are responsible. 

Reflecting these challenges, one theme that emerged for those organisations at lower levels of AI maturity was a tendency towards less nuanced approaches in building business cases and risk management. These organisations tended to use a default ‘yes’ or ‘no’ approach to the consideration of AI. Those inclined to ‘yes’ were driven by a desire to ‘keep up’ with rapidly evolving trends and competitors, while potentially downplaying or misunderstanding risks and the need for risk mitigation. At the other end, those inclined to ‘no’ focused on identifying and avoiding all potential risks associated with AI, consequently downplaying or ignoring potential benefits and opportunities. Neither approach is likely to deliver the right long-term outcome for the business or its customers. 

The siloed approach to AI and AI risk management within organisations increases the likelihood of governance and strategy shortcomings. The absence of a more multi-disciplinary approach connected to broader organisational practices and values increases the prospect of important risks being missed or underestimated and/or overconfidence in mitigating or managing those risks. In terms of business strategy, the absence of broader awareness and education in regard to AI capabilities and potential runs the risk of opportunities being overlooked.  

CEDA’s workshops also identified a significant disconnect in the prioritisation of trust and consumer confidence. While the vast majority of workshop participants (88 per cent) reported that trust and consumer confidence in AI is a high or very high priority for the future, only 12 per cent indicated this is a current priority or focus. The concern here is that practices developed today may not incorporate the governance foundations that enable safe, sustainable and responsible AI. If AI consumer confidence is not a current priority it won’t feature in corporate strategy considerations and important future opportunities may be lost. Retrofitting the right practices, processes, strategic thinking and culture is hard. More fundamentally, insufficient attention to these issues could result in bad outcomes in the early stages of AI implementation, which adversely impact confidence and trust and, in turn, AI opportunities now and into the future. More needs to be done to lift the priority and focus on building trust and confidence in AI. 

The still-evolving AI governance practices and challenges identified by most workshop participants are broadly consistent with those identified in the 2019 Department of Industry, Innovation and Science pilots. This suggests a need to step up efforts to support the effective design, development and implementation of safe, sustainable and responsible AI. 

• In response to this, CEDA will use its national convening power to: 
• Educate on the benefits of responsible AI;
• Showcase best practice and the positive impacts of responsible AI; and 
• Share practical learnings and case studies. 

In doing so, CEDA will seek to work collaboratively with other key stakeholders such as government, business and academia to magnify the impact and reach of this work and to identify further initiatives critical to Australia becoming a global leader in responsible AI.  

There is a clear appetite among CEDA members to understand what responsible AI means in practice for their organisations, and how to build and maintain capabilities, robust systems and frameworks as a basis for trust and confidence in the design and use of AI. CEDA will build and leverage member engagement to share expertise and learnings focused on: 

• Elevating the immediate priority of consumer trust and confidence as a way of driving a greater focus on the governance of AI more broadly;
• Enabling shared understanding of responsible principles for AI, what these mean in practice, and how to measure and report against them;
• Building multi-disciplinary whole-of-organisation approaches to the oversight and governance of AI; and
• Promoting the importance of building AI competencies and literacy, reflecting the ubiquity of digitisation, data collection and use.

In the absence of better AI business and governance practices, AI adoption will continue to lag in Australian organisations with long-term consequences for innovation, productivity and international competitiveness. This underscores the importance of learning from the challenges of implementing responsible AI and the need to accelerate efforts to improve education, understanding and the promulgation of best practice. 

Artificial intelligence (AI) is taking the world by storm and increasingly shaping economic and broader opportunities. It has become clear that AI has enormous potential to advance economic and societal wellbeing and enable improved environmental outcomes. 

AI is a collection of interrelated technologies used to solve problems autonomously and perform tasks to achieve defined objectives without explicit guidance from a human being. These technologies include machine learning, natural language processing, virtual assistants, robotic process automation, unique identity, video analytics and more.