{"slug": "show-hn-feedback-ama-meditionbench-llms-for-mediation-and-conflict-resolution", "title": "Show HN: Feedback/AMA MeditionBench; LLMs for mediation and conflict resolution", "summary": "Human Assisted Intelligence founder launched MeditionBench at mediationbench.com, a benchmark testing frontier LLMs on mediation and conflict resolution, with a research paper published alongside the results. The founder said Claude nearly saturated the current test, prompting a harder v4, and noted Astra scored surprisingly low, possibly due to alignment or safety efforts; Azure provided token credits for OpenAI and Anthropic models. The project aims to test whether AI can reduce polarization and improve how people communicate and reach agreement.", "body_md": "I am the founder of Human Assisted Intelligence and I believe AI can be used to reduce polarization and improve the way people communicate and reach agreement.\n\nAt [https://mediationbench.com/](https://mediationbench.com/), there are many frontier LLMs tested, but it is difficult to keep up. I've been trying to get more free openrouter or huggingface credits. \nAzure for OpenAI and Anthropic have both provided token credits. (Thank You!)\n\nThe research paper is there as well.\n\nThere have been efforts to use LLMs to lie and play diplomacy, which are excellent compliments to this work.\n\nIn v4, the test will get harder as Claude nearly saturated the test. Astra was surprisingly low, perhaps because of alignment or safety efforts.\n\nI'd love to get feedback and hear from all of you. I've been working on related ideas for 20 years, so it's definitely a passion project with too much time in isolation.\n\nComments URL: [https://news.ycombinator.com/item?id=50491841](https://news.ycombinator.com/item?id=50491841)\n\nPoints: 1\n\n# Comments: 0", "url": "https://wpnews.pro/news/show-hn-feedback-ama-meditionbench-llms-for-mediation-and-conflict-resolution", "canonical_source": "https://news.ycombinator.com/item?id=50491841", "published_at": "2026-11-13 23:50:03+00:00", "updated_at": "2026-11-14 00:20:33.928145+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-safety", "ai-research"], "entities": ["Human Assisted Intelligence", "MeditionBench", "Claude", "Astra", "OpenAI", "Anthropic", "Azure", "OpenRouter"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/show-hn-feedback-ama-meditionbench-llms-for-mediation-and-conflict-resolution", "markdown": "https://wpnews.pro/news/show-hn-feedback-ama-meditionbench-llms-for-mediation-and-conflict-resolution.md", "text": "https://wpnews.pro/news/show-hn-feedback-ama-meditionbench-llms-for-mediation-and-conflict-resolution.txt", "jsonld": "https://wpnews.pro/news/show-hn-feedback-ama-meditionbench-llms-for-mediation-and-conflict-resolution.jsonld"}}