The Romanian AI Factory infrastructure, being developed as part of a European project involving the National Institute for Research and Development in Informatics - ICI Bucharest, could become fully operational in 2027, the institute's Director General, Adrian Victor Vevera, said on Thursday at the 'Digital Finance in Central and Eastern Europe' Forum.

'I want to talk about our project, Romanian AI Factory. This is a European project through which we are beginning to build Romania's first AI infrastructure. The project has already been signed and work thereon is under way. The National Institute for Research and Development in Informatics is the host institution for this infrastructure. Together with the National University of Science and Technology Politehnica Bucharest and other partners, we are part of a consortium working to develop applications... We hope to have everything completed next year and the infrastructure fully ready for use,' Vevera said.

According to him, the project aims not only to develop the technological infrastructure but also to create an ecosystem for artificial intelligence research, training and use.

'We have already started building the ecosystem because, once this infrastructure is in place, we will be able to ensure the sovereignty of our data and run applications using confidential data. At the same time, we will be able to train people to use applications, develop applications and make better use of artificial intelligence in their day-to-day activities. We are part of the European network of artificial intelligence infrastructures,' the ICI Bucharest head explained.

Referring to the use of artificial intelligence in financial operations, he stressed that final decisions must remain the responsibility of humans and that autonomous systems must be capable of being shut down in situations involving risk.

'Artificial intelligence is a tool for us. We can give it a certain degree of autonomy and use it as a tool or an assistant, but the final decision must be made by a human. It is very important to maintain cybersecurity because we work with a great deal of confidential data. When developing artificial intelligence systems based on autonomous agents, we must ensure that there is a technological kill switch for situations in which something starts heading in the wrong direction,' Vevera said.

In his turn, Mihnea Alexandru Moisescu, Dean of the Faculty of Automatic Control and Computer Science of the National University of Science and Technology Politehnica Bucharest, warned that artificial intelligence systems are not yet ready to make fully autonomous decisions without significant risks.

'In my opinion, AI systems are not currently ready to make decisions on their own without a major risk of adversely affecting the processes in which they are involved. Critical thinking is very important, and education must focus on how people interact with AI and assess its decisions,' Moisescu explained.

As regards financial supervision, Paul Baranga, Director of the Financial Supervisory Authority's Integrated Supervision Directorate, said that AI-based tools are already being used in certain supervisory activities, while others are still being tested.

'At European level, artificial intelligence is being integrated into supervisory activities. These tools are used to process data from financial entities, analyse information published by companies and identify unusual market behaviour. Some of these tools have already been incorporated into day-to-day operational activities, while others are still at the experimental and testing stage,' the representative of the Financial Supervisory Authority said.

The 'Digital Finance in Central and Eastern Europe' Forum is organised by the Financial Supervisory Authority (ASF), in cooperation with the Institute of Financial Studies (ISF). AGERPRES (RO - writing by: George Banciulea; EN - writing by: Simona Klodnischi)