Polis is a platform that leverages machine intelligence to scale up deliberative processes. In this paper, we explore the opportunities and risks associated with applying Large Language Models (LLMs) towards challenges with facilitating, moderating and summarizing the results of Polis engagements. In particular, we demonstrate with pilot experiments using Anthropic’s Claude that LLMs can indeed augment human intelligence to help more efficiently run Polis conversations. In particular, we find that summarization capabilities enable categorically new methods with immense promise to empower the public in collective meaning-making exercises. And notably, LLM context limitations have a significant impact on insight and quality of these results.
However, these opportunities come with risks. We discuss some of these risks, as well as principles and techniques for characterizing and mitigating them, and the implications for other deliberative or political systems that may employ LLMs. Finally, we conclude with several open future research directions for augmenting tools like Polis with LLMs.
In particular, §2.2.4 concurs with your (and my) concerns about this post:
The existence of powerful vote prediction technology creates incentives to entirely replace human participation with in-silico deliberation. This runs the risk of amplifying existing biases as well as eliminating the many positive externalities of deliberation on mutual understanding, civic empowerment, surfacing leaders, etc. We strongly advocate that vote prediction be used only to amplify participants’ voices, never to replace them. We reject as invalid on both ethical and performance grounds the use of vote prediction technology to replace human participants with simulations.
We cannot enough emphasize the latter point: it would be catastrophic for deliberation at scale if the remarkable capabilities of LLMs lead to replacement of whole groups of individuals by simulacrums designed by a very different population. [emphasis in original] Besides the question of accuracy and biases, we believe this could also lead to a crisis of faith and belief in the deliberative process.
Opportunities and Risks of LLMs for Scalable Deliberation with Polis, a paper out this week, investigates the application of LLMs to assist human democratic deliberation:
In particular, §2.2.4 concurs with your (and my) concerns about this post: