This research explores how Large Language Models can revolutionize online discussions by moving beyond basic moderation to actively improving discourse quality.

• Beyond Content Moderation: LLMs can serve as discussion facilitators that enhance interaction quality, not just filter harmful content
• Dual Security Benefits: Addresses immediate threats (hate speech) while strengthening long-term democratic discourse
• Practical Applications: Offers frameworks for both evaluating discussion quality and designing intervention systems
• Emerging Capabilities: Highlights how recent LLM advancements enable more nuanced understanding of discussion dynamics

This research matters because it positions AI as a solution to both immediate security concerns around harmful content and longer-term threats to social cohesion and democratic values in digital spaces.

Evaluation and Facilitation of Online Discussions in the LLM Era: A Survey