Trustworthy AI
Appropriate use of AI offers significant economic and other benefits. However, AI innovation and adoption are hampered by a lack of trust. Building trust demands more than technical sophistication.
RAISE proposes a pipeline where application-driven research in the Technology and Frameworks pillars (TRL 2-4) is put into practice through the Systems pillar (TRL 5-7), actively demonstrating efficacy in the real world.
1. Trustworthy Technology
Developing technical advancements in transparency, robustness, explainability, security, and safety.
- Explainable and verifiable AI
- Transparency in medical AI
- Targeted GenAI models to mitigate hallucinations
- Ethical norms in agentic AI for public services
2. Trustworthy Frameworks
Our goal is to establish thought leadership in accountability, ethics, human rights, governance, and regulation.
- Shaping fit-for-purpose AI governance
- Aligning with international best practices (OECD, UNESCO, AI Safety Institutes)
- Embedding ethical and legal norms
3. Trustworthy Systems
Building infrastructure and solutions collaboratively with industry and the public sector.
- Sovereign Large Language Model (LLM)
- Privacy-preserving environmental monitoring
- Subsidised consulting for SMEs
Examples
Our priority projects include developing novel methods for explainable and verifiable AI to achieve transparency in medical AI diagnostic tools, and partnering Vision Mātauranga in an AI context.