How can artificial intelligence strengthen democracy and public trust? And what makes an AI system truly trustworthy?
These were the central questions at the CZDEMOS4AI conference, which brought together researchers, public institutions, technology experts and civil society to discuss the development of specialised AI assistants for areas such as crisis communication, media resilience, law and education.
One of the key ideas behind the project is that trustworthy AI should not only provide accurate answers, but also explain where those answers come from, communicate uncertainty openly and remain transparent throughout the decision-making process.
Transitions contributed to the discussion by bringing the perspective of media literacy, public trust and digital inclusion. We highlighted that resilience is not built solely by giving people access to reliable information—it also depends on whether they understand why they should trust a system and feel confident using it.
We also emphasised the importance of involving older adults and other underrepresented groups in the design and evaluation of AI tools. If these technologies are intended to serve society, they should be developed with the people who will ultimately use them.
The discussions also opened opportunities for future collaboration, particularly around user testing, focus groups and participatory approaches that can help ensure AI systems are not only technically robust, but also understandable, accessible and trusted in practice.
We are pleased to contribute this perspective to the ongoing development of CZDEMOS4AI and look forward to continuing the conversation on responsible, human-centred AI.