aiqbench analysis
This study reframes model collapse as a cross-contamination epidemic within the AI ecosystem, utilizing a bilayer SIR framework to demonstrate that immunity wanes and shared corpora remain susceptible to re-infection. The analysis identifies synthetic-text detection as the highest-leverage parameter for mitigation, suggesting that intervention strategies should prioritize detection-based filtering and multi-source data mixing to attenuate collapse dynamics. While agent-based models confirm these mean-field predictions for dense networks, the findings indicate that diversity loss persists even with modest source diversification under high contamination fractions.
Epidemiology of Model Collapse: Modeling Synthetic Data Contamination via Bilayer SIR Dynamics
arXiv:2606.05168v1 Announce Type: cross Abstract: Training on synthetic data causes model collapse, but existing analyses treat this as single-chain degradation. In reality, the AI ecosystem involves cross-contamination: models ingest synthetic data from other models, produce new synthetic text, and contaminate shared corpora. We propose a bilayer coupled SIR/SIRS framework -- a phenomenological mean-field model treating data corpora and AI models as two interacting populations, each with susceptible, infected, and recovered compartments linked by cross-layer transmission. The SIRS variant (our primary recommendation) incorporates immunity waning, reflecting that filtered corpora and retrained models remain susceptible to re-contamination. We derive the basic reproduction number $R_0 = \sqrt{\beta_D \beta_M / [(\gamma_D+\mu_D)(\gamma_M+\mu_M)]}$ via the Next Generation Matrix and apply standard epidemic threshold results to the bilayer system. Illustrative scenario-based calibration from public AI text prevalence data yields supercritical dynamics ($R_0 > 1$) across three scenarios; Sobol sensitivity analysis identifies synthetic-text detection as the highest-leverage parameter. A bipartite-network agent-based model confirms mean-field consistency ($R^2 > 0.96$) for dense networks but degrades under heterogeneity. GPT-2 contamination chain experiments (192 runs across WikiText and Shakespeare) show dose-response degradation and diversity loss qualitatively consistent with the threshold picture. Matched-budget source-diversity experiments (1,088 runs) provide suggestive evidence that multi-source mixing modestly attenuates collapse, but the effect vanishes at lower contamination fractions. Intervention analysis identifies detection-based filtering and herd immunity as the highest-leverage strategies.
Abstract:Training on synthetic data causes model collapse, but existing analyses treat this as single-chain degradation. In reality, the AI ecosystem involves cross-contamination: models ingest synthetic data from other models, produce new synthetic text, and contaminate shared corpora. We propose a bilayer coupled SIR/SIRS framework -- a phenomenological mean-field model treating data corpora and AI models as two interacting populations, each with susceptible, infected, and recovered compartments linked by cross-layer transmission. The SIRS variant (our primary recommendation) incorporates immunity waning, reflecting that filtered corpora and retrained models remain susceptible to re-contamination. We derive the basic reproduction number $R_0 = \sqrt{\beta_D \beta_M / [(\gamma_D+\mu_D)(\gamma_M+\mu_M)]}$ via the Next Generation Matrix and apply standard epidemic threshold results to the bilayer system. Illustrative scenario-based calibration from public AI text prevalence data yields supercritical dynamics ($R_0 > 1$) across three scenarios; Sobol sensitivity analysis identifies synthetic-text detection as the highest-leverage parameter. A bipartite-network agent-based model confirms mean-field consistency ($R^2 > 0.96$) for dense networks but degrades under heterogeneity. GPT-2 contamination chain experiments (192 runs across WikiText and Shakespeare) show dose-response degradation and diversity loss qualitatively consistent with the threshold picture. Matched-budget source-diversity experiments (1,088 runs) provide suggestive evidence that multi-source mixing modestly attenuates collapse, but the effect vanishes at lower contamination fractions. Intervention analysis identifies detection-based filtering and herd immunity as the highest-leverage strategies.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2606.05168 [cs.CL]
(or arXiv:2606.05168v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2606.05168 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 317 words
Related coverage
The Fairness Collapse Phenomenon: Bias Amplification in Language Models Trained on Synthetic Data
arXiv · Aug 6, 2026
SynthSentry: Detecting Synthetic Data Contamination in Language Model Training Data
arXiv · Sep 14, 2026
The Economics of Model Collapse: Equilibrium, Welfare, and Optimal Provenance Subsidies in Synthetic Data Markets
arXiv · Aug 19, 2026
Distribution-First Population Simulation: Collapse, Calibration, and Recall in Non-WEIRD LLM Persona Modeling
arXiv · Jul 23, 2026