Artificial intelligence is advancing rapidly and becoming increasingly widespread in industry and public services. In demanding applications such as industrial equipment control and fault diagnostics, it must be possible to trust the decisions made by AI. The international CLEAR project (Comprehensive Learning for Enhanced AI Responsiveness) is developing cost-effective and transparent solutions that make it possible to justify and verify AI behaviour, even in challenging environments.
Industrial processes and critical infrastructures continuously generate large volumes of diverse multimodal data of varying quality. Current AI models are not always able to make reliable use of information originating from different sources, and the reasoning behind their decisions may remain unclear. This makes it difficult to deploy AI in environments where operational reliability and safety are essential.
The CLEAR project develops methods for combining sensor data, images, documents and other data types into one coherent and understandable entity. At the same time, the project explores how AI decision-making can be made more transparent through approaches such as fine-tuned models, methods based on verified data sources, and energy-efficient smaller models. The project also develops intelligent agents that can independently carry out specific tasks and support people in decision-making.
“Companies need AI which can genuinely be trusted. In the CLEAR project, we are developing solutions that not only predict what is likely to happen but also explain clearly why the AI generates a particular outcome,” says Principal Scientist Johannes Peltola from VTT. “When the basis for decision-making is transparent, AI becomes a truly useful tool for critical industrial and societal needs.”
CLEAR is an international project within the ITEA4 programme. In Finland, the project is partially funded by Business Finland. The international consortium includes organisations from Sweden, Belgium, Canada, Estonia, Germany, Portugal, Spain and Türkiye.
The Finnish participants in the project are Fastems, Nitor, Vaadin, VTT, Wapice and Y4Works. In Finland, the work focuses on agent-based systems, the development of AI models suitable for industrial use, improving the trustworthiness of AI, and end-user applications. The aim is to support customers’ strategic planning, sales and software development, while also improving productivity through, for example, situational awareness based on sensor data in a range of industrial, energy-intensive and critical infrastructure environments.
The CLEAR project is expected to deliver significant benefits for Finnish industry and society more broadly. When data can be used more effectively and AI-generated decisions are more transparent, processes become more reliable and safer. At the same time, energy-efficient and cost-effective solutions strengthen companies’ competitiveness in rapidly evolving international markets.
