ml-model-collapse-synthetic-data

IN premise — entries/2026/06/21/wiki-Machine_learning-chunk-5.md

Created 2026-06-21T09:55:51+00:00

Model collapse is the degradation that occurs when models train on uncurated synthetic data or outputs of prior model versions, also called 'model autophagy disorder (MAD)'

Dependents

These beliefs depend on this one:

• IN attack-surface-expands-at-every-ml-boundary — ML systems face adversarial threats across multiple stages — adversarial examples exploit model inference (perturbations cause misclassification across supervised and reinforcement learning), data poisoning targets the training pipeline (injected false data degrades model performance), and model collapse can degrade models trained on uncurated synthetic data — suggesting a broad attack surface where vulnerabilities span different phases of the ML lifecycle.
• OUT federated-learning-resolves-privacy-data-tradeoff — Federated learning would resolve the fundamental tension between data access and privacy in ML — decentralized training preserves privacy by keeping raw data on user devices, while synthetic data generation (GANs) provides unlimited augmentation without real patient data, together enabling ML development without compromising individual privacy.
• OUT generative-models-provide-unlimited-training-data — Generative models (GANs, VAEs, diffusion) would provide effectively unlimited synthetic training data — generating privacy-preserving medical images, augmenting scarce datasets, and enabling training without data collection barriers — fundamentally resolving the labeled-data bottleneck that constrains supervised learning.
• IN model-collapse-recursive-crisis-amplifier — Model collapse from synthetic data creates a recursive amplifier within ML's compounding reliability crisis — as capable models generate training data for next-generation models, reliability degradation is inherited and compounded across model generations, meaning capability scaling now directly poisons the data substrate on which future capability depends, adding a temporal feedback dimension to the crisis.
• OUT modern-llm-pipeline-sustainable — Modern LLM training pipelines (self-supervised pretraining → instruction tuning → RLHF) are a sustainable methodology — they dissolve classical paradigm boundaries by successfully combining all three ML paradigms, and self-supervised learning provides an effectively unlimited source of training signal.
• OUT synthetic-data-safe-replacement-for-real-data — Synthetic data from generative models can safely replace real training data at scale, enabling privacy-preserving ML pipelines and unlimited data augmentation without degradation.