In the realm of artificial intelligence (AI) governance, a fascinating debate emerges: Can our current political systems effectively govern AI, or is there a need for a radical shift in political participation? The crux of the matter lies in the question of whether politicians or ordinary citizens should be the ones steering the course of AI development and regulation. This is where the ideas of political theorist Hélène Landemore and democratic innovator Peter MacLeod come into play, offering a compelling case for a more inclusive and participatory approach to AI governance.
The Limitations of Traditional Politics
Landemore and MacLeod argue that our current political systems are not equipped to handle the complexities of AI governance. They point out the declining health of political parties, with collapsing memberships and eroding trust in government. This has led to a system that is overly responsive to the affluent, as measured consistently over the last 40 years in political science studies. The selection mechanism for legislators, they argue, is partly at fault, favoring socio-economic elites and perpetuating an oligarchic system.
The Power of Citizens' Assemblies
MacLeod has spent over 20 years putting the idea of citizens' assemblies into practice in Canada. These assemblies, he explains, are like juries, with letters sent out to thousands of households inviting them to spend up to 12 days examining a single topic. The key difference is that assembly members work towards finding consensus, not just pushing their own agendas. This process fosters a sense of solidarity and self-worth, as people feel recognized and affirmed, and come to feel a greater and expanded sense of their own self-worth. It's a powerful tool for creating a more democratic and inclusive society.
Landemore describes these assemblies as being filled with love, a striking way to describe politics. She attributes this to the fact that people feel like they matter and count, something that is often lacking in their daily lives. This sense of solidarity and love is crucial for creating a more democratic and inclusive society, and it's a key reason why citizens' assemblies could be a path forward for AI governance.
The Role of Citizens in AI Governance
The question of AI governance has stumped politicians, regulators, and the tech industry alike. Landemore argues that citizens' assemblies could crack this problem. She suggests that citizens are good at dealing with trade-offs, and could be asked questions like: How much are you willing to pay in additional electricity costs, in depletion of natural resources, in the increased risks of war over rare minerals for the convenience of using AI tools like ChatGPT every day? MacLeod agrees, suggesting that citizens' assemblies could foster a greater adherence to the precautionary principle, and a more conservative and guarded approach to AI development.
The Limitations of Synthetic Publics
Some argue that AI could replace human deliberation entirely with so-called 'synthetic publics'. MacLeod disagrees, pointing out that the sense of solidarity and love that emerges in citizens' assemblies is crucial for creating a more democratic and inclusive society. Landemore adds that the change of preferences that happens in deliberation is not just in reaction to good arguments, but also a reaction to the love and solidarity felt for the people with whom one deliberates. Automating everything to an agentic AI that takes preferences as given would not be able to synthesize this solidarity, because we are human creatures.
In conclusion, the debate over AI governance highlights the need for a more inclusive and participatory approach to politics. Citizens' assemblies offer a compelling solution, fostering a sense of solidarity and self-worth, and providing a platform for citizens to engage in the complex questions of AI development and regulation. While the challenges are significant, the potential benefits are immense, and the time is ripe for a radical shift in political participation.
