Problems and Purpose

AI technology brings benefits and conveniences but also could make the problem of misinformation worse. At the same time, AI technology can also help us identify information sources and analyze the authenticity of information.

Institute of Science, Technology and Society at National Yang Chiao Tung University (NYCU), commissioned by the Ministry of Digital Affairs (Moda), conducted the deliberative poll on utilizing AI to enhance information integrity on March 23 in 2024 to analyze the public views, attitudes and positions on policy issues. The opinions of the citizens and digital practitioners after the deliberation could help establish more comprehensive AI evaluation standards, ensuring that the development of AI meets social expectations.

The citizen deliberative assembly provided an open and equal platform for discussion, allowing citizens from all over the country to discuss how large platforms should use AI technology to enhance information integrity. Two deliberative sessions included four discussion topics: Should large platforms enhance self-regulation, or should government regulation be strengthened? How should large platforms use AI to help citizens analyze emerging threats to information integrity? How should large platforms use AI to help users maintain information integrity? How should large platforms use AI to enhance information integrity standards? Multiple policy proposals were designed for the four main discussion topics, allowing participating citizens to comprehensively review, weigh, and discuss different options.

Background History and Context

Taiwan officially became a partner of the international non-governmental organization, the Collective Intelligence Project (CIP) in 2023. Moda initiated the Alignment Assemblies project, collaborating with international organizations to provide the people with more reliable AI systems. Also, Moda collaborated with the Industrial Technology Research Institute to set up the AI Product and System Evaluation Center in December 2023, targeting language models for its initial assessment.

Institute of Science, Technology and Society at National Yang Chiao Tung University (NYCU) is devoted to deliberative democracy and enhancing scholarly understanding of the interdependence of science, technology and society, raising social awareness and offering insights to policymakers. Institute of Science, Technology and Society collaborated with Deliberative Democracy Lab at Stanford University and conducted the deliberative polling on Taiwan’s Pathway to 2050 Net-Zero Emissions in 2022 and “Responsible AI Innovation” in 2023.

Organizing, Supporting, and Funding Entities

The online deliberative assembly was organized by Institute of Science, Technology and Society at NYCU in collaboration with the Deliberative Democracy Lab at Stanford University. The project was commissioned by the Ministry of Digital Affairs and Administration for Digital Industries and Industrial Technology Research Institute.

Participant Recruitment and Selection

Recruitment involved sending survey through the official “111” SMS platform to 200,000 randomly generated mobile numbers, resulting in a total of 1,760 valid respondents. On March 23^rd, the entire afternoon featured two sessions of deliberation assemblies attended by a total of 447 citizens, who were randomly assigned to 44 citizen deliberation groups. Among these, 12 groups consisted of digital practitioners and 32 groups were composed of general citizens. A total of 436 participants attended the entire duration of the citizen deliberation assembly and completed both pre- and post-deliberation survey.

Methods and Tools Used

This project adopted the deliberative polling® method, randomly recruiting over four hundred citizens from across the nation, including digital practitioners, to gather on the Stanford Online Platform to discuss how to utilize AI to enhance information integrity.

What Went On: Process, Interaction, and Participation

The project produced two short videos (each about 4 and a half minutes) introducing the discussion topics, which were played before the discussions in the citizen groups to help stimulate ideas and focus the discussions on the issues outlined in the briefing materials. During the online deliberations, the platform provided information related to the discussion topics, available for participants to review and facilitate structured discussions. After thorough discussions, each citizen group proposed two questions they most wanted to ask experts, which were then addressed by interdisciplinary AI scholars during the plenary session.

Influence, Outcomes, and Effects

Results show that citizens perceived the discussion topics as more complex than initially anticipated, enhancing their understanding of the challenges in AI technology development after deliberation. About 85% citizen participants support stronger self-regulation by large platforms. 89% participants indicate that large platforms should employ new technologies to ascertain whether certain platform content is AI-generated, marking a 2.6 percentage point increase. There was also a slight increase in support for enhancing the information integrity in evaluations of AI products and systems. Citizens showed improved accuracy on knowledge questions after deliberation and expressed positive views on the citizens’ deliberation assembly.

Analysis and Lessons Learned

The citizen groups provided a wide range of opinions, reflecting diverse perspectives and highlighting the profound reflections and concerns about AI ethics, social responsibility, and public governance in the context of the rapid development and application of AI technology. Three key aspects and recommendations from citizen groups include: 1) Enhancement of AI regulatory framework and standards; 2) Improving platform capabilities in AI for analysis and identification of information; 3) Establishment and refinement of evaluation mechanisms.