Abstract
In the aftermath of crises, from world wars to global pandemics, governments often seek resilience in their rebuilding efforts. Prioritising resilience as an organisational value entails developing decision-making processes that are stable and adaptable.
Technology can help, but in recent decades governments have tended to use technology to further other values, such as economic efficiency. Modern data-intensive technologies, such as data science and artificial intelligence (AI), hold tremendous potential to rebuild resilience in government. However, changes are needed in order to realise this potential.
Researchers from the AI for science and government (ASG) programme at The Alan Turing Institute have been addressing these challenges under the multidisciplinary theme of ‘Shocks and resilience’ (S&R). In this white paper, we argue that developing resilience requires a new and distinctly public sector approach to data science, in which data-intensive technologies do not just automate or replicate what humans can already do well, but rather do things which people cannot – such as tackling difficult, multi-sector problems that have no ‘right’ solution. We illustrate our argument with selected case studies based on projects involving ASG researchers.
Based on our experiences under the S&R theme and within the broader ASG programme, we propose five recommendations for building resilience into government using data science:
- Provide ethical guardrails for data science in government.
- Invest in ready-to-go data infrastructure and models.
- Distil essential causal mechanisms from complex systems.
- Utilise collective modelling approaches.
- Work across boundaries to share insights between domains.
Following these recommendations will place governments in a better position to tackle the growing list of existential problems that loom, from the next pandemic to global environmental collapse.