{"slug": "steelagent-llm-orchestrated-physics-informed-steel-property-prediction-with-and", "title": "SteelAgent: LLM-Orchestrated, Physics-Informed Steel Property Prediction with Audited Generalization and Calibrated Uncertainty", "summary": "A study posted to arXiv (2611.06953v1) reports that SteelAgent, an LLM-orchestrated system coupling metallurgical descriptors and machine-learning models for steel property prediction, uncertainty reporting and inverse design, shows that held-out unit choice changes conclusions drawn from SteelBench tensile-strength evaluations. For gradient boosting, pooled out-of-fold R² fell from 0.859 on a random split to 0.113 when steel families were held out, and empirical coverage of nominal 90% prediction intervals dropped from 91.0% under random splitting to 37.6% on held-out sources. Physics-derived features raised gradient-boosting R² by 0.2534 in a 17-family holdout analysis but lowered it by 0.358 in a separate four-grade laboratory analysis, leading the authors to recommend reporting the evaluation population alongside any property estimate or compliance score.", "body_md": "arXiv:2611.06953v1 Announce Type: new \nAbstract: Steel property models can perform well on random test splits even when they extrapolate poorly to unfamiliar alloys. SteelAgent couples metallurgical descriptors and machine-learning models with a language-model interface for property prediction, uncertainty reporting and inverse design. We examine how the choice of held-out unit changes the conclusions drawn from SteelBench tensile-strength evaluations. For gradient boosting, pooled out-of-fold $R^2$ fell from 0.859 on a random split to 0.113 when steel families were held out. Empirical coverage of nominal 90% prediction intervals ranged from 91.0% under random splitting to 37.6% on held-out sources. The effect of physics-derived features depended on the evaluation. They increased gradient-boosting $R^2$ by 0.2534 in a 17-family holdout analysis and decreased it by 0.358 in a separate four-grade laboratory analysis. We report group-level errors and a synthetic regression experiment illustrating how calibration can deteriorate on held-out groups. These results motivate reporting the evaluation population alongside a property estimate or compliance score.", "url": "https://wpnews.pro/news/steelagent-llm-orchestrated-physics-informed-steel-property-prediction-with-and", "canonical_source": "https://arxiv.org/abs/2611.06953", "published_at": "2026-11-06 05:00:00+00:00", "updated_at": "2026-11-06 05:20:08.463829+00:00", "lang": "en", "topics": ["machine-learning", "ai-research", "ai-agents", "large-language-models"], "entities": ["SteelAgent", "SteelBench", "arXiv"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/steelagent-llm-orchestrated-physics-informed-steel-property-prediction-with-and", "markdown": "https://wpnews.pro/news/steelagent-llm-orchestrated-physics-informed-steel-property-prediction-with-and.md", "text": "https://wpnews.pro/news/steelagent-llm-orchestrated-physics-informed-steel-property-prediction-with-and.txt", "jsonld": "https://wpnews.pro/news/steelagent-llm-orchestrated-physics-informed-steel-property-prediction-with-and.jsonld"}}