Information resilience
Multiple industry reports and recent media coverage identified ‘data gone wrong’ as the biggest risk factor for AI and other emerging technologies, and its impact is increasingly recognised as a threat from mundane cybercriminals to sophisticated well-funded entities, raising concerns of national security.
Information resilience is the capacity of organisations to build, protect, and sustain agile data pipelines, capable of detecting and responding to failures and risks across the value chain in which the data is sourced, shared, transformed, analysed, and consumed.
The success of AI’s implementation will require robust mechanisms and capacity building around information resilience. Prof. Shazia Sadiq leads research to address these issues and includes:
- Responsible use of data assets (principled approaches to data governance, access and sharing)
- Data curation at scale (machine learning, crowd-sourcing and human-in-the-loop techniques)
- Algorithmic transparency (approaches to promote interpretability, uncertainty quantification unbiasedness, transparency and reproducibility in the design of learning algorithms)
- Trusted data partnerships (data literacy, trust in data linking, lifting barriers to data sharing)
- Creating value from data (data monetisation, business process improvement, measuring analytics value and organisational structures).
