Lithuania does not currently have an artificial intelligence (AI) transparency standard to help understand what algorithmic tools are used by public sector institutions and for what purposes. Overall, the country lacks a strategic vision, clear commitments and a responsible authority to mitigate the risks and maximize the benefits of AI technologies, according to the research conducted by Transparency International Lithuania.
Public sector institutions in Lithuania are currently using algorithmic tools to improve the efficiency of internal processes, monitoring and public service delivery. At least 1 in 4 institutions (12 out of 53) do so. Fully automated decision-making, where an algorithm makes the final decision without human intervention, has not yet been implemented in the public sector.
According to the participants of this research, some of the main risks associated with the use of algorithmic tools are bias, lack of testing, the use of bad datasets, or the fact that algorithmic tools might not be adapted to the country context. The biggest benefit currently identified is the reduction of mechanical, routine work.
According to the official inquiries, 4 out of 12 institutions (Ministry of Defence, National Courts Administration, Public Procurement Office, State Tax Inspectorate) are continuously monitoring algorithms to ensure that the decisions made are accurate and ethical.
Public sector employees lack knowledge about the use and implementation of AI, which can put the quality of decision-making at risk. For example, there is currently insufficient investment in enhancing employee skills, fostering knowledge on monitoring the impact of AI, and providing guidance on preparing technical specifications when procuring such tools from suppliers.
“Artificial intelligence is inevitably making its way into decision-making in the public sector. That is why I would very much like to see our government representatives take the lead and clearly agree on a strategy for the use of AI in Lithuania and a standard for transparency in AI. We need to know who is behind the decisions that directly affect our lives and how well-informed they are”, said Ingrida Kalinauskienė, CEO of Transparency International Lithuania.
“The Netherlands Government has been one of the first to use algorithms in decision making. But we have learned the hard lesson that technologies may bring speed, but not necessarily accuracy. To avoid bias we need to keep the human element involved to remain true to our standards”, – H.E. Ambassador of the Kingdom of the Netherlands Jack Twiss Quarles van Ufford.
Taking into account the main insights and risks identified in the study, TI Lithuania proposes:
- For the Ministry of Economy and Innovation to actively engage and promote a nationwide discussion on the strategic development and the use of artificial intelligence within the country. Simultaneously, convene a professional community to discuss and agree upon the national AI strategy, the transposition of the EU AI Act, the creation of AI transparency standard, and the establishment or designation of a responsible institution for AI in Lithuania.
- For the institution responsible for the implementation of AI strategic goals to regularly collect and publish summarised statistics on the use of AI technologies in the public sector in Lithuania, including information on the specific AI technologies used, their purposes, and the allocated funding.
- Develop an AI transparency standard, which would ensure the public disclosure of information related to the use of AI in each public sector institution. This standard should be structured in an open data format, including details such as algorithm name, description, supplier information, and channels for reporting discrepancies. Contracting authorities should implement the AI transparency standard when formulating technical specifications and determining contract performance conditions for public procurement.
- For public sector institutions to enhance the competence and knowledge of their employees to create and/or acquire purposeful AI technologies. Employees should be equipped to use these solutions effectively, evaluate their impact, and understand the resulting changes.
- For public sector institutions that use algorithmic tools conduct regular assessments of AI systems to ensure accuracy and ethical operation. They should consider establishing mechanisms for collecting feedback and implementing improvements.
The study’s insights and recommendations were prepared based on 7 semi-structured in-depth qualitative interviews conducted by TI Lithuania between September and October 2023 with representatives of public sector institutions, academia, as well as data from official inquiries (55) sent in June 2023.
For more information: Ingrida Kalinauskienė, email@example.com, +370 5 212 69 51
The initiative is implemented in cooperation with the Embassy of the Kingdom of the Netherlands in Lithuania.