The application of data science and use of AI systems in government should enable better decisions by politicians, policymakers, and communities themselves. However, the decision landscape in government is characterised by complexity, ambiguity, and the politics of policy formation processes. Far from popular belief, useful information and evidence are often disparate and scarce; data privacy and other human rights are paramount; the cost of a poor decision is high and often irreversible; there are many and varied stakeholders all with different priorities, and the benefits and costs of digital technologies have the potential to be inequitably distributed.

As a result of these characteristics many communities and governments around the world face decision paralysis, confounded by the exponential growth of data supplies, insufficient data analysis capabilities, and inconsistent use of evidence in policymaking.

If we are to make progress towards the global Sustainable Development Goals (SDG), for example, we need to design and implement policy programs that allow for the complexity of dynamic intersecting causal pathways across themes of societal and planetary wellbeing. This in turn requires adaptive experimentation that can respond to different scales of responsibility and levels of impact and decision systems that are polycentric, occurring at overlapping organisation levels from individual communities to local government to private sector and national policy and regulation. To address these issues needs a framework that can;

• Identify and structure shared questions that matter most to inform the decision-making needs in policy design and implementation processes.
• Understand the decision-making systems and needs, informing collective decision points to increase value.
• Distinguish between cause and effect – to prioritise policy design and identify what works and impacts outcomes.
• Adaptively uncover what information is needed to answer the question at hand, using real-time experimental design and thus measuring what matters most.
• Acknowledge and explain estimates of the uncertainty inherent in inferring these cause-and effect relationships to reinforce robust and explainable policy decisions that build trust with the public.
• Fuse together many disparate types of information. Information from lived experience, from existing relevant data, from expert opinion and from bespoke data collection.
• Co-create and co-design with impacted communities to ensure the interpretation and adoption of recommendations in ways that are meaningful to local culture and context.
• Establish governance procedures that ensure AI systems are accurate, accountable, fair, and fit for purpose.

There is no one discipline that can address all these issues. The call for submissions to this special track session will focus on the intersection of these disciplinary knowledge sets driven around the specific high-stakes problems including, but not limited to, breaking entrenched cycles of disadvantage, environmental change, and democratic resilience. It will highlight the necessity for multi-disciplinary research areas and the skills necessary to solve them.

Successful submissions may tackle any specific government issue but should include at least two of research sub-categories and have a defined government policy high-stakes application.

Subcategory: Methods for causal inference and estimation

• Randomized control trials for government decision-making.
• Learning the structure and associated parameters of Directed Acyclical Graphs (DAGs) and the impact of interventions on DAGs.
• Causal inference from longitudinal studies

Subcategory: Collective Intelligence techniques for policy formulation.

• Question science to measure what matters.
• Techniques for combining qualitative and quantitative information.
• Platforms, systems, and management practices that facilitate the emergence of collective intelligence.
• Methods for establishing, explaining, and communicating confidence in AI outputs for government decision-making

Subcategory: Translation of analytic outputs findings into policy and community practice.

• Implementation science for the co-creation and adoption of policy recommendations in the community. • Value of data through a chain analysis.
• To establish the uptake of research findings into government policy-making processes including data capability and data science capacity.

Subcategory: Governance systems and accountability for integration into public policy.

• Regulatory and policy approaches for AI governance for government agencies.
• Governance mechanisms, assurance, and assessment framework.