{
  "abstract": "Closed-loop human-AI systems generate high-dimensional behavioural trajectories whose collective dynamics remain obscure. Using 297,915 learners' adaptive-tutoring histories, we define semantic order variables before model fitting and test them in user-disjoint cohorts. The state exhibits reproducible basin-like flow and operationally defined, state-heterogeneous metastable-like kinetics. A construction-matched null distinguishes normalised-memory relaxation from a reproducible excess field. A four-term conditional mechanism recovers population drift (r = 0.946; learner-bootstrap 95% CI, 0.935-0.955). Predictive event-level self-supervised learning recovers the state and learned-plane flow; null-referenced corrections retain directional, partial-amplitude excess-field structure without full calibration. Shuffled-order training reverses learned-plane flow on ordered trajectories; support-alignment randomisation selectively reduces inward transport. Both axes remain linearly accessible without state supervision. Without cross-model fitting, the models share leading population drift (r = 0.866; learner-bootstrap 95% CI, 0.857-0.875) and persistence ordering; residual directions remain model-specific. These results identify an externally anchored leading-order effective field linking empirical dynamics, an interpretable mechanism and neural computation.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2608.30946",
  "author_total": 5,
  "authors": [
    {
      "name": "Minlin Wu"
    },
    {
      "name": "Xu Fang"
    },
    {
      "name": "Yicheng Zhang"
    },
    {
      "name": "Chenyu Zhou"
    },
    {
      "name": "Zhiyi Liu"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/f69yq8nhjkp1n.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2608.30946",
  "fulltext_url": null,
  "id": "f69yq8nhjkp1n",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2608.30946",
  "language": null,
  "licence": "LicenseRef-arXiv-nonexclusive-1.0",
  "licence_label": "arXiv licence",
  "licence_source": "article",
  "licence_url": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "nlin.AO"
    ],
    [
      "openalex",
      "1702|Intelligent Tutoring Systems and Adaptive Learning"
    ]
  ],
  "openalex": "W7204915698",
  "pdf_url": "https://arxiv.org/pdf/2608.30946",
  "pmcid": null,
  "pmid": null,
  "published": "2026-08-31",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/f69yq8nhjkp1n.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:17:41Z",
      "native_id": "2608.30946",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2608.30946"
    }
  ],
  "subjects": [
    {
      "field": "physics",
      "origin": "arxiv:nlin.AO",
      "subfield": "statistical-nonlinear"
    }
  ],
  "title": "Reproducible macroscopic dynamics in a closed-loop human-AI learning system",
  "title_text": "Reproducible macroscopic dynamics in a closed-loop human-AI learning system",
  "type": "preprint",
  "updated_at": "2026-10-30T12:17:41Z",
  "url": "https://research.draconitic.com/paper/f69yq8nhjkp1n",
  "venue": "arXiv",
  "versions": []
}