{"slug": "firstresearch-auditable-question-formation-for-llm-scientific-discovery-agents", "title": "FirstResearch: Auditable Question Formation for LLM Scientific Discovery Agents", "summary": "Researchers introduced FirstResearch, a framework that generates auditable research questions for LLM scientific discovery agents by producing a structured Research Question Certificate. In evaluations, FirstResearch outperformed baseline methods, scoring 4.86/5 versus 4.38/5 for the strongest baseline, with certificate removal causing scores to drop below 1/5. The framework aims to make LLM-generated scientific questions more inspectable by exposing assumptions, mechanisms, and falsifiable hypotheses.", "body_md": "arXiv:2607.05682v1 Announce Type: new\nAbstract: LLM systems for scientific discovery increasingly assist with ideation, literature synthesis, experiment planning, and report generation, but the first research question they propose can remain difficult to audit: it may sound plausible without exposing the mechanism, falsifier, or assumption that a scientist should inspect. We introduce FirstResearch, a first-principles research-question formation framework for scientific LLM agents whose core artifact is a structured Research Question Certificate. The certificate records primitive definitions, assumptions, a mechanism model, a tension or contradiction, a falsifiable hypothesis, a minimal decisive test, and a failure update rule, making the proposed question inspectable before downstream execution. On ten LLM-agent research topics, FirstResearch outperforms controlled prompt-level baselines inspired by AI co-scientist, Agent Laboratory, and AI Scientist-v2 under a primary DeepSeek-blind-judge protocol. A Gemini-2.5-Flash independent-judge rescore of the same 40 baseline packages preserves the system-level ranking, with FirstResearch scoring 4.86/5 versus 4.38/5 for the strongest baseline and Pearson agreement of 0.865 on average score. A one-repeat ablation checkpoint further suggests that the certificate-centered core is the strongest component: certificate-only scoring reaches 4.90/5 under DeepSeek and 4.88/5 under Gemini, while removing certificates drops below 1/5 under both judges. These results are preliminary and use LLM judges rather than human domain experts, but they support a narrow scientific-discovery claim: explicit derivation constraints are a promising mechanism for making LLM-generated scientific questions more auditable. Code, prompts, saved outputs, and reproduction scripts are available at https://github.com/louiswang524/FirstResearch.", "url": "https://wpnews.pro/news/firstresearch-auditable-question-formation-for-llm-scientific-discovery-agents", "canonical_source": "https://arxiv.org/abs/2607.05682", "published_at": "2026-07-08 04:00:00+00:00", "updated_at": "2026-07-08 04:04:32.046710+00:00", "lang": "en", "topics": ["large-language-models", "ai-research", "ai-safety", "ai-agents"], "entities": ["FirstResearch", "DeepSeek", "Gemini-2.5-Flash", "AI Scientist-v2", "Agent Laboratory", "AI co-scientist"], "alternates": {"html": "https://wpnews.pro/news/firstresearch-auditable-question-formation-for-llm-scientific-discovery-agents", "markdown": "https://wpnews.pro/news/firstresearch-auditable-question-formation-for-llm-scientific-discovery-agents.md", "text": "https://wpnews.pro/news/firstresearch-auditable-question-formation-for-llm-scientific-discovery-agents.txt", "jsonld": "https://wpnews.pro/news/firstresearch-auditable-question-formation-for-llm-scientific-discovery-agents.jsonld"}}