Login
iscover
As Artificial Intelligence (AI) continues to evolve, ensuring its deployment is safe, responsible, and trustworthy is paramount.
Artificial Intelligence (AI) is revolutionising the defence sector, offering unprecedented opportunities to enhance decision-making, improve efficiency, and elevate warfighter safety and capability. However, as AI continues to evolve, ensuring its deployment is safe, responsible, and trustworthy is paramount.
Global Approaches to Trustworthy AI
The European Union (EU) and the United States (US) are pivotal to the future of global AI governance. Both regions recognise AI's transformative potential and the need to manage associated risks. The EU has taken a rules-driven approach with its AI Act, categorising AI systems based on risk levels and imposing specific obligations on high-risk systems. This approach, combined with existing regulations like the General Data Protection Regulation (GDPR), ensures that AI systems respect data privacy and human rights.
In contrast, the US currently has adopted a more flexible, principles-based approach, as outlined in the Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence. This encourages bottom-up innovation while promoting fairness, transparency, and accountability, though it lacks the stringent centralised and federal enforcement mechanisms pursued by the EU. Both approaches emphasise the importance of trustworthy AI, a principle that Deloitte integrates into its AI solutions.
Trustworthy AI in Defence
The US’s military AI initiatives, particularly under the AUKUS alliance, will significantly influence Australia's defence capabilities. In November 2023, the US Department of Defense's (US DoD) Chief Digital and Artificial Intelligence Office (CDAO) released the Responsible Artificial Intelligence (RAI) Toolkit. This toolkit, a major component of the US DoD's Responsible AI Strategy & Implementation Pathway, offers tools and guidance for developing, deploying, and using AI systems responsibly. The RAI Toolkit currently includes approximately 70 open-source, industry-standard tools designed to support various RAI-related activities, with a strong emphasis on fairness, transparency, and accountability in AI.
Approach for Australia
Reflecting the recent release of the National AI Assurance Framework for the Australian government, Deloitte recommends that the Australian Department of Defence (AUS DoD) develop a specialised adaptation of the framework akin to the CDAO approach. This adaptation should include robust guidelines and standards tailored to the unique operational needs of military applications, aligning with both national and international standards. It should emphasise operational integrity, resilience under adversarial conditions, and strict adherence to ethical norms. Furthermore, as outlined in the National AI framework, the Australian government's role as an exemplar in the safe and responsible use of AI is vital. The AUS DoD is ideally positioned to lead this initiative, showcasing best practices in AI implementation that ensure lawfulness, ethical integrity, and technical robustness. This strategy not only supports the evolution of AI capabilities within a secure and responsible regulatory framework but also fosters innovation and safety. Risk-based AI approaches for Defence aim to balance fostering innovation with ensuring the responsible and ethical use of AI systems. This approach allows for the prioritisation of risk management and compliance resources by both AI actors and regulators on higher-risk applications while minimising undue burdens on low-risk AI systems.
Optimal Reality: Enhancing Defence Operations
Deloitte's AI and digital twin solution, Optimal Reality (OR), exemplifies how advanced AI technologies can enhance mission-critical operations by augmenting human capabilities. OR, currently deployed across various sectors and clients, uses sophisticated AI algorithms to analyse vast amounts of data, providing operators with actionable insights for better-informed decisions. It can identify potential risks across various scenarios, equipping operators with the necessary information to take preemptive measures and reduce the likelihood of adverse events. OR uses a variety of techniques (e.g. machine learning, neural networks, physics-based simulation) to achieve the appropriate levels of transparency, interpretability and scalability necessary for a wide range of AI use cases.
Deloitte sees immense value in integrating OR into Australia’s security operations. This integration would enable the Australian Defence Force (ADF) to achieve a strategic advantage through enhanced decision-making, improved situational understanding, increased efficiency, safety, and robust scenario planning. By aligning with comprehensive AI regulations and frameworks, Deloitte, through OR, can showcase Australia's sovereign capability to design, develop, and implement AI that is safe, responsible, and trustworthy.
Conclusion
Defence industry will continue to develop AI technologies, with or without AUS DoD guidance. Guidance from AUS DoD is vital to ensure that AI technologies are developed and implemented in a way that aligns with Australia’s national interests, safety standards, and ethical considerations. By adopting a risk-based regulatory framework, defence leaders can harness AI's benefits while ensuring its deployment is safe, responsible, and trustworthy.
Deloitte can assist in developing AI regulation, policy and frameworks. Additionally, Deloitte can further demonstrate how AI compliance is achievable through the implementation of Deloitte's OR for defence and national security use cases. Leveraging advanced AI capabilities like Deloitte's OR will provide significant military advantages, ensuring that mission-focused technology adapts to evolving threats, operational demands, and strategic objectives.
To find out more, visit www.deloitte.com/au/defencecapability
Authored by Sibi Ravindran, Director AI & Data, Deloitte
