Agentic AI – Physicist Collaboration in Experimental Particle Physics: A Proof-of-Concept Measurement with LEP Open Data

Anthony Badea, Yi Chen, Marcello Maggi, Yen-Jie Lee, Electron-Positron Alliance · 2026

Read Agentic AI – Physicist Collaboration in Experimental Particle Physics: A Proof-of-Concept Measurement with LEP Open Data by Anthony Badea, Yi Chen, Marcello Maggi, Yen-Jie Lee, Electron-Positron Alliance (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-05735.txt.

Introduction

Badea and team present an agent thrust-spread measure in electron-positron collisions at 91.2 GeV using ALEPH data. The work proceeds through agent-physics teamwork prompting Claude and Codex. Under physics guidance, agents wrote and ran all code and drafted most text. A fully fixed thrust spectrum comes from repeat Bayesian opening and Monte Carlo fixes. The main point is showing agent analysis with thrust as a benchmark. The work steps toward a theory-test loop where agents assist measures and math and blend insights by matching results.

Catalog record

CREATORAnthony Badea, Yi Chen, Marcello Maggi, Yen-Jie Lee, Electron-Positron Alliance
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE109852
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.05735

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Agentic AI – Physicist Collaboration in Experimental Particle Physics: A Proof-of-Concept Measurement with LEP Open Data Anthony Badea, Yi Chen, Marcello Maggi, Yen-Jie Lee, Electron-Positron Alliance Abstract We present an AI agentic measurement of the thrust distribution in e + ​ e − e^{+}e^{-} collisions at s = 91.2 \sqrt{s}=91.2 GeV using archived ALEPH data. The analysis and all note writing is carried out entirely by AI agents (OpenAI Codex and Anthropic Claude) under expert physicist direction. A fully corrected spectrum is obtained via Iterative Bayesian Unfolding and Monte Carlo based corrections. This work represents a step toward a theory–experiment loop in which AI agents assist with experimental measurements and theoretical calculations, and synthesize insights by comparing the results, thereby accelerating the cycle that drives discovery in fundamental physics. Our work suggests that precision physics, leveraging the open LEP data and advanced theoretical landscape, provides an ideal testing ground for developing advanced AI systems for scientific applications. 1 Context of e^{+}e^{-} Alliance Achieving Standard Model predictions at the level of precision required for future collider programs, including the High-Luminosity Large Hadron Collider (HL-LHC), is essential to ensure robust control of systematic uncertainties in the interpretation of potential new physics anomalies. The demand for such precision is driven by percent-level experimental uncertainties and the unprecedented statistical power of modern collider datasets. In this regime, theoretical uncertainties become limiting, and the dominant contribution arises from our modeling of Quantum Chromodynamics, often referred to as precision QCD. Precision QCD studies can be pursued at hadron colliders such as the LHC, but they are most cleanly and systematically tested in lepton collider environments. At present, however, there is no active high-energy lepton collider operating at LEP energies or above. To address this, the e^{+}e^{-} alliance 27 has successfully curated and reanalyzed archived datasets from the Large Electron–Positron collider (LEP), the precursor to the LHC, opening new opportunities enabled by three decades of theoretical advances and modern experimental techniques. The combination of ultra-precise e^{+}e^{-} data and state-of-the-art perturbative and non-perturbative QCD tools provides a uniquely powerful environment for high-precision studies. The alliance has produced a series of precision measurements using these archived datasets, including two-particle correlations 5 , jet energy spectra and substructure 13 , long-range near-side correlations 14 , energy-energy correlators 8 , and unbinned measurements of thrust 4 . More broadly, this program is a compelling opportunity for AI for Science in a domain where precision is paramount. While AI is increasingly used to assist theoretical calculations and generate predictions, in the physical sciences such predictions ultimately require experimental validation. In high-energy physics, this validation depends on collider datasets that are rarely publicly accessible. The e^{+}e^{-} environment, particularly at LEP energies, provides the cleanest setting for precision tests of QED and QCD, with well-defined initial states and calculable dynamics. The successful curation and use of archived open data therefore establishes the conditions for a theory–experiment loop in which AI systems, guided by domain experts, can propose new calculations, observables, or improved models and directly confront them with high-precision data. This creates a concrete framework for developing AI systems that do not merely interpolate existing knowledge, but actively refine theoretical descriptions through direct comparison with experiment. Lessons from AI–Physicist Collaboration Carrying out this analysis has also produced practical insights into how AI agents and physicists can work together effectively. Our workflow followed a constrained human-in-the-loop cycle in which the physicist defined physics objectives, acceptance criteria, and interpretation authority, while the AI agent owned implementation throughput, consistency checks, and traceability. Crucially, experienced physicists also contributed domain intuition that cannot be reduced to formal specifications alone, including judgments about physically meaningful binning, acceptable regularization behavior, uncertainty communication, and when apparent anomalies warranted deeper investigation. Requests were made concrete and executable, each iteration linked to specific scripts, outputs, and note sections, and corrections were applied incrementally and revalidated rather than batched into large unreviewed changes. From this experience, several recurring challenge classes emerged. Implicit physics priors, expert expectations about acceptable binning, regularization behavior, or uncertainty presentation, were often assumed rather than explicitly stated, requiring iterative clarification. Precision formatting constraints (label offsets, legend placement, pad geometry) required many short feedback loops, and apparent cosmetic changes sometimes affected interpretability and therefore physics conclusions. State synchronization across rapidly evolving scripts, outputs, and note text demanded continuous provenance refreshing. Throughout, authority remained intentionally asymmetric: the physicist owns physics choices and release signoff, while the agent executes and refactors. These lessons motivate stricter templating for figures, stronger configuration contracts, and clearer channels for encoding expert heuristics into machine readable constraints in future iterations of this program. 2 Introduction Machine learning has been central to high-energy physics for decades, enabling reconstruction algorithms, event classification, and detector optimization across collider experiments. A new frontier has emerged with large language models (LLMs) developed by major AI laboratories, including Anthropic 3 , OpenAI 24 , Google DeepMind 17 , and others, that can now assist directly with scientific reasoning, mathematical proof, and code generation. A striking asymmetry persists: AI has transformed the theoretical and computational side of fundamental physics research—assisting derivations of new results in theoretical particle physics 25 , designing novel protein structures for biology and health research 16 , and simplifying complex calculations. Yet the experimental side of the AI-assisted research loop remains largely absent. To realize a future where AI rapidly accelerates the discovery of new physics, AI-assisted scientific workflows must be developed and validated across both precision theory and precision experiment. A logical first step toward closing this gap is an agentic analysis of real experimental data with expert physicist oversight. This is precisely the approach pursued here: the physicist opened AI Agent (OpenAI Codex and Anthropic Claude Code), brainstormed the analysis goals, then iteratively prompted the AI agents to execute the full analysis workflow. No analysis code was written by hand; the physicist’s role is analogous to advising a graduate student, providing expert direction, reviewing results, and iterating. The agent was supplied with context comprising the published ALEPH thrust paper 20 , prior ROOT-based ALEPH analysis macros covering event selection and thrust calculation, internal analysis notes documenting the miniDST data format and detector calibration procedures, and the LEP open-data files themselves. No pre-written unfolding or systematic-uncertainty code was provided; instead, the physicist described the physics objectives and acceptance criteria in iterative prompts, and the agent constructed the corresponding implementations from those descriptions. This prompting strategy follows from the established understanding in the AI literature that few-shot examples and supplied context improve AI-assisted outputs 9 ; 29 . A detailed inventory of the inputs and prompting approach is given in Sec. 3 . Pursuing this theory–experimental interface for physics will inevitably deepen understanding of AI capabilities and their potential to improve AI systems for science domains where precision matters. This vision parallels the evolution of AI in neighboring scientific domains. In materials science and structural biology, AI work has progressed from computational prediction with post-hoc human validation toward AI-designed experiments executed by autonomous robotic systems, creating closed feedback loops where AI proposes, executes, measures, and learns. High-energy physics faces a parallel opportunity. While we cannot deploy robotic particle detectors (yet), the emergence of open, curated data, partially enabled by the e^{+}e^{-} alliance

Read the complete text