{"slug": "beyond-the-hivemind-escaping-llm-homogeneity-via-meta-persona-anchoring-and", "title": "Beyond the Hivemind: Escaping LLM Homogeneity via Meta-Persona Anchoring and Sequential Temperature Scaling", "summary": "Researchers propose Meta-Persona Anchoring with Filtered Temperature Scaling (FTS) to reduce the 'Artificial Hivemind' effect in large language models, cutting average pairwise cosine similarity from approximately 0.85 to 0.65 on the INFINITY-CHAT dataset using open-weight models under 20B parameters. The method uses a two-stage generation process with self-selected personas and extreme temperature scaling (T ≥ 4.0) to increase output diversity.", "body_md": "arXiv:2608.02618v1 Announce Type: new\nAbstract: Recent studies have identified an ``Artificial Hivemind'' effect in Large Language Models (LLMs) causing models to converge on a narrow, homogenized consensus even for open questions. This semantic collapse limits the diversity of AI, resulting in high inter-response similarity ($\\approx 0.80-0.90$) even under high-temperature sampling. In this paper, we propose a novel mitigation framework to increase diversity: Meta-Persona Anchoring combined with Filtered Temperature Scaling (FTS). Our approach utilizes a two-stage generation process: first, the model is prompted to self-select a unique, idiosyncratic persona to anchor its starting point; second, we apply a dual-stage sampling sieve, utilizing Top-$p$ filtering to preserve grammatical validity followed by extreme temperature scaling ($T \\ge 4.0$) on the surviving candidates to explore the broadened probability distribution. We evaluate our method using the INFINITY-CHAT dataset on state-of-the-art open weight models under $\\sim$20B parameters. Our results demonstrate a significant reduction in semantic convergence, with average pairwise cosine similarity dropping from ($\\approx 0.85$) to ($\\approx 0.65$). Our scheme achieves a majority of questions below the 0.7 threshold, effectively reducing the gap between artificial mode collapse and human-level typological diversity. We provide our implementation as an open-source framework to enable more diverse and creative AI deployments.", "url": "https://wpnews.pro/news/beyond-the-hivemind-escaping-llm-homogeneity-via-meta-persona-anchoring-and", "canonical_source": "https://arxiv.org/abs/2608.02618", "published_at": "2026-08-05 04:00:00+00:00", "updated_at": "2026-08-05 04:09:09.814469+00:00", "lang": "en", "topics": ["large-language-models", "generative-ai", "ai-research"], "entities": ["arXiv", "INFINITY-CHAT"], "alternates": {"html": "https://wpnews.pro/news/beyond-the-hivemind-escaping-llm-homogeneity-via-meta-persona-anchoring-and", "markdown": "https://wpnews.pro/news/beyond-the-hivemind-escaping-llm-homogeneity-via-meta-persona-anchoring-and.md", "text": "https://wpnews.pro/news/beyond-the-hivemind-escaping-llm-homogeneity-via-meta-persona-anchoring-and.txt", "jsonld": "https://wpnews.pro/news/beyond-the-hivemind-escaping-llm-homogeneity-via-meta-persona-anchoring-and.jsonld"}}