The "Meet the Future of AI" conference, held at the Belgian Institute for Postal Services and Telecommunications (BIPT) in Brussels on June 19th, 2024, brought together leading voices in the realm of generative AI and democracy. Organised by a cluster of European Research and Innovation projects - TITAN, AI4TRUST, vera.ai, AICODEPROJECT, AI4Media, and AI4Debunk - the event was a hub of discussion and debate about how generative AI can be both a threat and a tool in the fight against disinformation. The conference was divided into two key sessions, each featuring prominent speakers from various fields.
Opening the event was Krisztina Stump, Head of Unit, Media Convergence and Social Media, who emphasised the crucial role of technology in both the propagation and mitigation of disinformation.
Highlighting findings from the Eurobarometer, she pointed out that citizens across Europe consider disinformation the biggest threat to democracy. She stressed that while AI is often seen as a threat, it also offers significant opportunities to combat disinformation.
Stump highlighted the importance of ongoing projects under the H2020 and Horizon Europe programs, which have seen direct investments of €28 million.
Panel 1: Generative AI in Latest Elections and Disinformation Campaigns
Denis Teyssou from Agence France-Presse introduced the VeraAI and InVID-WeVerify plugins, tools designed to detect deepfakes and other forms of synthetic media. He discussed the evolving landscape of deepfakes, including face swapping and voice cloning, and the geographical variations in the techniques and intentions behind these fakes.
Aqsa Farooq, Co-chair of EDMO's working group on AI, presented preliminary findings from a survey on AI-generated disinformation during the recent European Parliament elections. Her research revealed that while AI-generated disinformation incidents were noted, they constituted a relatively small proportion of total disinformation. Farooq's survey also found significant concern among citizens about AI's role in elections, particularly in Poland.
Samuel Power from the University of Sussex explored how AI can enhance electoral oversight. He detailed a process for using OCR to analyse electoral spending and discussed the challenges and potential solutions for implementing such technologies within the UK Electoral Commission.
Tjade Stroband, Director of AI and Disinformation Policy for Microsoft in Brussels, outlined Microsoft's approach to ensuring electoral integrity through responsible AI. He introduced Microsoft's Copilot feature, emphasising the importance of transparency, safety by design, and AI literacy. Stroband also highlighted Microsoft's election protection tools and public awareness campaigns aimed at educating citizens about deepfakes.
Max van Drunen from the AI, Media, and Democracy Lab at UVA discussed the regulatory landscape surrounding generative AI and disinformation. He emphasized the importance of transparency, labelling misleading AI content, and ensuring reliable information is accessible.
Panel 2:
Harnessing Generative AI Against Disinformation and for Trust Building
Rasmus Nielsen from the Reuters Institute for the Study of Journalism at the University of Oxford discussed public expectations of generative AI in news and politics. He highlighted the paradox of disclosure, where the public expects transparency about AI use in news production but remains wary of its overall trustworthiness.
Riccardo Gallotti from FBK presented a study on the persuasive power of large language models (LLMs) in online debates, revealing that humans tend to be more convinced by AI, especially when debates are personalised.
Eliška Pírková from Access Now addressed concerns about AI-generated disinformation in the EU elections, suggesting that while the threat was smaller than anticipated, enforcement of existing regulations remains critical.
Francesco Saverio Nucci from (our) TITAN project explored how generative AI can enhance citizens' critical thinking. He proposed the idea of a personalized LLM chatbot coach to help users better identify disinformation through Socratic questioning.
A Q&A session after each panel featured insightful discussions on the challenges and opportunities of using AI to combat disinformation. Speakers highlighted the need for interdisciplinary collaboration and the importance of ongoing research to keep pace with technological advancements.
In closing the event, project officer Peter Freiss emphasized the necessity of bridging the gap between stakeholders from various sectors to address the complex challenges posed by generative AI. As a key takeaway, the event underscored the importance of actionable steps and collaborative efforts to leverage AI in the fight against disinformation, ensuring a healthier democratic process.
The Meet the Future 2024 conference was a significant step forward in understanding and harnessing the potential of AI in democracy, providing a platform for sharing knowledge, strategies, and innovations that could shape the future of electoral integrity and public trust.
Comments