Any attempt to regulate artificial intelligence is likely to be ineffective without first ensuring the availability of trusted large-scale sovereign data sets.
For the Australian government, AI presents transformative potential, promising to revolutionise the way in which government departments and agencies operate. The allure of AI-driven efficiency, precision and insight is irresistible. Yet, amid the chorus of AI evangelists, a discordant note rings true: establishment of robust AI policy guardrails now would be premature and potentially counterproductive without first addressing the fundamental issue of sovereign trusted data.
This contentious stance is rooted in the understanding that trustworthy AI hinges on the availability of trusted large-scale data sets. Without this bedrock, current attempts to regulate AI could end up being built on quicksand and become ineffective in mitigating the menace of misinformation and disinformation. Proposed Australian regulations are focusing on privacy requirements, labelling of AI-generated work, the legal consequences of AI choices and understanding how the software makes decisions.
AI, in its essence, is a data-driven phenomenon. The algorithms that power AI systems are not imbued with inherent intelligence; rather, they learn and evolve through ingestion and analysis of vast quantities of information. The quality, accuracy and representativeness of this data directly influence the performance and trustworthiness of the resulting AI models. In the absence of robust, verifiable data, AI can purvey misinformation, amplifying biases, perpetuating stereotypes and undermining public trust. Such risks are particularly acute for government agencies, for which stakes are high and the impact of erroneous decisions can be far-reaching.
Regulations that are not grounded in the realities of data quality and provenance risk being toothless and easily circumvented by those seeking to exploit AI for nefarious purposes.
Australian sovereign data is data that is owned, controlled and governed within Australia’s borders. It is subject to Australian laws and regulations, ensuring its collection, storage and use adhere to the highest standards of privacy, security and ethics. This control is crucial in mitigating the risks of foreign interference, data manipulation and the spread of misinformation. By maintaining sovereignty over the source data, the Australian government can ensure that the AI systems it deploys are built on a foundation of trust and transparency.
Sovereign data empowers Australian government agencies to build AI models that are tailored to the unique needs and context of the nation. By training AI systems on data that accurately reflects the diversity and complexity of Australian society, we can ensure that these models are not only effective but also equitable and just. Furthermore, sovereign data fosters transparency and accountability, allowing for independent scrutiny of the data and algorithms that underpin AI decision making. This transparency is essential for building public trust in AI and ensuring its responsible use in government.
Establishing trusted large-scale sovereign data sets is undeniably complex. It requires overcoming four challenges:
—Data Collection and Integration. Gathering comprehensive, high-quality data from disparate sources across government agencies is a logistical and technical challenge. Data must be standardised, cleaned and de-identified to ensure its usability and protect privacy.
—Data Governance. Robust data governance frameworks must be established to ensure data quality, security and ethical use. This includes defining clear roles and responsibilities for data management, implementing access controls and establishing mechanisms for addressing data breaches and misuse.
—Expertise and Resources. Building and maintaining sovereign data capabilities will require significant investment in infrastructure, technology and skilled people. Data scientists, analysts and governance experts are essential for ensuring effective management and utilisation of sovereign data.
—Cultural Shift. A cultural shift towards data-sharing and collaboration is needed across government agencies. Breaking down silos and fostering a culture of open data can accelerate the creation of comprehensive, multi-dimensional data sets that reflect the complexity of real-world challenges.
Despite these complexities, several countries have made significant strides in establishing sovereign data capabilities. The European Union’s General Data Protection Regulation (GDPR) is a prime example, setting a global standard for data privacy and control. India’s push for data localisation and its efforts to build a national digital infrastructure also highlight the growing recognition of the strategic importance of sovereign data.
The concept of sovereign data linkages with trusted nations also presents an opportunity for Australia. By establishing secure and mutually beneficial data-sharing agreements with like-minded and aligned countries, Australia could expand its access to high-quality data while maintaining control and ensuring ethical use. Such linkages would require careful negotiation and robust governance frameworks to ensure data privacy, security and alignment with shared values.
The Chinese government’s unfettered access to vast amounts of citizen data, coupled with its willingness to deploy AI for surveillance and social control, raises serious concerns about the future of AI ethics and governance. Australian collaboration with like-minded nations can counterbalance China’s AI ambitions.
Responsible and effective deployment of AI in Australian government is not merely a technical challenge but a strategic imperative. Sophisticated sovereign data sets can provide the bedrock for trustworthy AI, mitigating the risks of misinformation and disinformation while unlocking the full potential of AI for public good.