A new Norwegian research centre led by SINTEF and NTNU is shifting the focus of artificial intelligence from large language models like ChatGPT to small, specialized systems capable of autonomous decision-making. Funded by a substantial portion of Norway’s AI billion initiative, the centre aims to develop AI that supports critical decisions in sectors such as health, energy, logistics, and infrastructure. Research manager Signe Riemer-Sørensen emphasizes that current AI models merely simulate conversation, while future AI should act—making decisions and managing processes. Examples include AI systems that calculate insulin doses based on anticipated activity rather than static measurements, and energy models that optimize solar power usage to prevent grid overload.
The centre advocates for modular AI—many small models that handle specific tasks and collaborate. Professor Sebastien Gros highlights autonomy as central, with AI systems capable of deciding when to act or request assistance. The centre involves over 50 partners from industry and public sectors, including Equinor, Statkraft, Microsoft Norway, and the Norwegian Public Roads Administration. These partners seek both advisory systems and fully autonomous solutions.
The initiative also explores how AI can interpret sensor data and maintain trust in automated decisions. Despite varied applications—from snowplough routing to hospital triage—the underlying AI principles remain consistent. The centre’s work represents a paradigm shift: from generalist, language-based AI to task-specific, decision-oriented intelligence designed to integrate seamlessly into real-world operations.
Key Points
- SINTEF and NTNU lead a new AI centre focused on decision-making AI.
- Funded by Norway’s AI billion initiative (NOK 1 billion over five years).
- Emphasis on small, specialized AI models over large language models.
- Applications include health (e.g., insulin dosing), energy (e.g., solar grid management), logistics, and infrastructure.
- AI autonomy is central—systems must decide independently or seek help.
- Over 50 partners involved, including Microsoft Norway, Equinor, and Statkraft.
- AI models will interpret sensor data and support trusted decisions.
- Modular AI architecture enables collaboration between specialized models.
- Real-world examples include snowplough routing and hospital triage.
- The centre promotes cross-sector learning and shared AI principles.
Norway’s new AI centre, led by SINTEF and NTNU, pioneers a shift from conversational AI to autonomous, specialized models designed to make real-world decisions across health, energy, and infrastructure.
