"Three ideas to adress challenges of generative AI" - Prof. Urs Gasser at a panel discussion on sustainable and democratic development of GenAI

• Launch of a series of EU Grand Challenges to incentivize democracy-enhancing uses of Gen AI, supplementing the current focus of measures on developers and tool development. The “Grand Challenges” would be funded through an independently administered GenAI innovation fund, to which tech firms - starting with those calling for a moratorium - contribute voluntarily (like Swiss-based Migros is putting 1% of the annual turnover into cultural activities) or via taxes. Open source and open data as well as communities like Wikipedia, among others, would play a driving role, in collaboration with academia.

• Expand the mandate and funding of government-based digital innovation labs, including GovLabs, Data labs, Reallabore, etc., to develop and experiment with democracy-enhancing use cases in the public sector to reduce information asymmetries and improve communications between citizens and government. Trust issues in public discourse aren’t limited to misinformation in the context of elections but also involve the daily communications of core institutions of democracy, for instance in key areas such as access to justice, explaining government policy, access to resources, etc. Strategic collaborations with networks like the TUM Venture Labs and similar efforts could further boost social entrepreneurship at this critical juncture.

• To address serious challenges ahead involving more pervasive forms and novel distribution channels of mis- and disinformation campaigns involving Generative AI, build a networked infrastructure inspired by CERTs in the field of cybersecurity, bringing together information and media quality experts and practitioners from the public and private sectors. The network would establish real-time alert systems (building upon incident databases), self- and mutual aid mechanisms, and coordinated response schemes, in addition to educational campaigns. Existing AI Observatories (like OECD), the Global Network of Internet & Society Centers, civil society expert networks, and the Euro Tech Universities Alliance could serve as a backbone at the incubation state. The AI act would need to be amended to allow for access to data, models, and parameters (inspired by the DSA).