A. National and global challenges
AC1: AI in healthcare and medicine
AI in healthcare and medicine seeks to test and validate results from RAs to ensure smoother and more efficient patient flow and improved diagnostics/treatment. The activities will focus on AI-enabled digital twins for: (i) more efficient bed allocation in and across hospital wards and municipal health services, including prediction of the need for health workers; (ii) surgical interventional support, with focus on anomaly detection, and (iii) leveraging large longitudinal multimodal datasets about patients, to train and test AI models for diagnostics/treatment of cancer and cardiovascular diseases.
AC2. AI in the mobility system
AI in the mobility system addresses existing multimodal mobility systems (physical/digital infrastructure, traffic and fleet management, in-vehicle functionality), and next generation mobility systems (with smart infrastructure, fleets of cooperative and automated vehicles equipped with real-time learning AI- systems). Utilising high quality transport data in Norway, we aim to: (i) make long-term planning and prediction of maintenance more accurate, sustainable, and cost efficient; (ii) achieve collective awareness and optimal movement of vehicles/people interacting with the infrastructure; and (iii) investigate safety in next generation AI-guided mobility ecosystems.
AC3.AI for case processing and management in the public and private sector
AI for case processing and management in the public and private sector seeks to address the Auditor-General’s critique in 2024 that Norway has been slow in using AI for this purpose. We will (i) demonstrate our new methods to semi-automate decision-making and adjust them to accommodate legal standards (GDPR, AI domain) and concerns over bias; (ii) use regulatory sandboxes, new data infrastructures, and qualitative fieldwork with citizens and marginalised groups to test the trustworthiness in practice of such automation; and (iii) develop common digital platforms and proposals for legislative reform.
AC4. AI for enhanced planning and operation of physical/digital infrastructure
AI for enhanced planning and operation of physical/digital infrastructure aims at optimal allocation and use of resources in buildings, transport, offshore, electricity grids, ports, ships. TRUST will test: (i) integration of planning, construction, and maintenance of infrastructure with green AI-solutions; and (ii) new AI-functionality and governance schemes for cost-efficient data mining, high precision maintenance, and removal of negative environmental impacts from physical infrastructures.
AC5. AI for monitoring violations in armed conflict
AI for monitoring violations in armed conflict seeks to develop AI capabilities for humanitarian law and peace/ceasefire agreements. Open-source digital intelligence, satellite imagery, drone footage, and CCTV cameras – together with decentralised citizen engagement – enable real-time analyses. TRUST will test: (i) veridical methods for assembling large, complex, and incomplete datasets in conflict zones; (ii) integration of legal standards in automated violations analysis through knowledge-informed AI: (iii) use of temporal-spatial AI models to improve analysis of factual sequencing; and (iv) adaptive/inclusive governance approaches on public control of data.
AC6. Combatting criminality with AI
Combatting criminality with AI seeks to enable data aggregation, behaviour prediction, and selected real-time AI responses to AI-driven criminality, while addressing concerns about privacy, explainability, security, and fairness (e.g., predictive policing and discrimination). We will test: (i) temporal-spatial foundation and anomaly detection models on financial and surveillance data to create digital evidence; (ii) the efficacy and sociological effects of responsible UQ and causal explainability in policing settings; and (iii) how to balance privacy and discrimination for high-risk AI in practice.
AC7. AI for adaptation to a new climate and nature regime
AI for adaptation to a new climate and nature regime covers applications for energy efficiency, environmental protection, health challenges, geohazard preparedness, and emissions reduction. We will test (i) AI-based quantification of socio-economic benefits of climate adaption; (ii) AI-guided monitoring of fair use and ownership of biological resources; (iii) AI-guided preparedness for creating reliable, localised predictions of extreme events (land-slides, glacier instability, heat waves, etc.); and (iv) climate impacts on health and social stability in under-resourced communities.
B. Innovation in technology, governance, and science
AC8. National and international governance for trustworthy AI
National and international governance for trustworthy AI demonstrates the capacity of our solutions to reach trustworthy governance at scale and over time. This will include: (i) testing a Norwegian governance feedback loop between AI deployers and AI supervisory authorities for anticipatory governance; (ii) piloting a governance incubator for ‘one-stop shop’ analysis of legal obstacles crossing multiple partners; and (iii) identifying selected catalytic spaces to embed future-proof governance mechanisms, through selected legislative triggers, economic incentives, and targeted information campaigns.
AC9. Assurance of AI-systems
Assurance of AI-systems involves verifying its accuracy, robustness, and compliance with ethical and regulatory principles and standards. Activities include: (i) achieving a common understanding across sectors and users of the technical, legal, and ethical challenges in the AI assurance processes; (ii) test how assurance should be conducted during implementation and use of AI in selected cases.
AC10. AI for time-series analysis and anomaly detection
AI for time-series analysis and anomaly detection will be demonstrated in several domains. We will test (i) our results on forecasting energy production/demand and safety diagnostics at production sites (offshore wind, ship ecosystems, financial systems, surgical theatres) and (ii) methods in online learning and AI-guided decision support systems based on multimodal data.
AC11. AI-guided decision making
AI-guided decision making with qualified recommendations is operationally important in the private and public sector. TRUST will test (i) whether selected proposed systems for AI-guided decision making are trustworthy and assist flourishing in practice; and (ii) whether these conditions hold if these cognitive companions are transferred across institution/domain and scaled up.
AC12. AI for enhancing security and preparedness
AI for enhancing security and preparedness will test how new trustworthy AI systems detect, analyse, and respond to security incidents. For preparedness to anticipated/unexpected events, we will integrate into surveillance and security systems new methods for (i) embedded security and anomaly detection, (ii) assessing AI robustness, and (iii) AI prediction of accidents and risks.
AC13. Green AI for the green transition
Green AI for the green transition covers developing AI solutions that both are intrinsically green and promote the green transition. We will test out: (i) AI-systems at the edge, using compositional digital twins; and (ii) AI-based decision making and optimisation in production and distribution of intermittent renewable energy (offshore wind).
AC14. Life-long learning and AI
Life-long learning and AI will address how knowledge and competence development for individuals and organisations ensure AI-driven disruption leads to positive societal transformation. We will: (i) test which learning processes, and forms of integration of AI tools, foster trustworthy uses of AI through user-oriented studies with partners; and (ii) develop a program for learning, which will be a sustainable mechanism for the partnership between the academia and national/international organisations.
AC15. AI for sciences
AI for sciences involves the demonstration of trustworthy AI in scientific research. We will focus on evaluating the influence of our novel AI methodologies and algorithms on the discovery processes across multiple scientific fields. We anticipate contributing to discoveries in our ongoing work in biomedicine, geosciences, physics, psychology, and law, for example in climate research, personalised cancer treatment, predictive power of AI in legal reasoning, and memorability studies in neuroscience.