AI and the Future of Democracy
Clarote & AI4Media / Better Images of AI / Power/Profit / CC-BY 4.0
By Gordon Johnstone (Communications and Events Manager at the Scottish AI Alliance)
Podcast Episode: AI and the Future of Democracy with Prof Gina Neff
2024: the year the world votes
In 2024 alone, there are 64 countries scheduled to hold elections, representing approximately 49% of the world’s population.[1] If my LinkedIn is to be trusted, approximately 100% of people are talking about AI. When you combine these societal events, you get a heady cocktail of hyperbolic discourse around how AI is going to impact democracy – not to mention a thousand generated images of Donald Trump with a six-pack. AI isn’t just generative art and memes, though. The risks it poses to the democratic process are well-documented and oft-cited[2]. For an in-depth analysis of these risks, you can peruse this paper from the Northeastern University Ethics Institute: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4805026
AI and Amsterdam: a ‘dam good time
On the more positive side of the debate, there are some very exciting developments at the intersection of politics and AI that are likely to play a role in all our lives in the future. Last month I was delighted to attend the University of Amsterdam’s AI, Media and Democracy Lab’s conference, titled ‘AI’s Impact on Society, Media and Democracy’. This two-day event brought together academics and practitioners from around the world to discuss the far-reaching applications artificial intelligence systems may offer the democratic process. While there was far too much to adequately recap in a short blog, a few points in particular have stuck with me.
In his talk ‘Towards Hyperdemocracy’, Professor Takayuki Ito of Kyoto University highlighted the potential for large-scale online discussion platforms moderated by AI agents to revolutionise political discourse. Currently, large scale citizen platforms are very difficult to moderate and manage by human agents due to the rapid, complex, and often confusing, way online conversations unfold. By deploying AI agents into citizen platforms, his team were able to facilitate creative agreements and consensus at previously impossible scales – the agents could act quicker than any human ever could. This is an exciting step towards improving democratic participation and empowering marginalised voices in society. By carefully automating moderation of online discussions we can ensure that it’s not just those who shout the loudest who dominate political discourse, while also amplifying thoughtful points that may have flown under the radar.
Later that day, I listened to Michiel Bakker talk about how his team at DeepMind are training large language models as ‘deliberative assistants’ that analyse discussions and generate statements which attempt to reflect a consensus between the participants. His research found that these assistants helped virtual citizens’ assemblies discuss and find agreement on controversial topics. Dr Mark Klein and his team at the MIT Center for Collective Intelligence are using AI-powered software tools to assist large-scale discussions reach ‘pareto-centric’ decisions, where the tools are used to recognise and prioritise the decisions and consensuses that benefit the greatest number of people.
The common theme running through these applications of AI is the facilitation of fruitful discussion and consensus building. Anyone who has spent more than a few minutes on X will recognise both of these things as desperately needed in online political discourse. It was also heartening to see that every presentation showed ethics to be at the heart of the research being conducted. It’s very easy (and often quite fun) to get bogged down in the concerns around AI and lose sight of the fact that innovative, empowering work is being done around the world to make our democracies stronger and more resilient. Events like the one in Amsterdam are a wonderful reminder of that.
The Future of Democracy (in podcast form)
Learning how to spot AI-generated content – especially in a political context – is going to be an important skill to develop over the coming months and years. The Guardian recently posted a useful guide to spotting deepfakes and gen AI – you can read it here: https://www.theguardian.com/technology/article/2024/jul/01/seven-signs-deepfake-artificial-intelligence-videos-photographs
Upon returning from Amsterdam, I was lucky enough to record a podcast with Dr Gina Neff – Executive Director of the Minderoo Centre for Technology and Democracy at the University of Cambridge – about AI and the upcoming elections. From the erosion of our shared reality to the reinvigoration of political participation, Dr Neff’s insights are absolutely fascinating. If you’re concerned about AI in the political system, I’d recommend giving it a listen and staying informed in the latest developments in AI technology. You can listen to the podcast anywhere you find your podcasts, or by clicking here: https://pods.link/SAIApodcast
[1] https://time.com/6550920/world-elections-2024/
[2] https://www.politico.eu/article/deepfakes-distrust-disinformation-welcome-ai-election-2024/