MIT CSAIL AI Artist Unlearning & Attribution Lab

Research Model: Style Latent Space Indeterminacy

Experimental Setup

MIT CSAIL Core Insight: Pruning 100% of an artist's direct training data leaves stylistic synthesis intact due to proxy visual features distributed across adjacent datasets.

Latent Style Manifold (D3 Simulation)

Training Points Active
Style Retention
100.0%
Attribution Entropy
1.18
Manifold Drift
0.000

Synthesis Probe & Attribution Trace

Baseline Synthesis
Post-Pruning Output
Multi-Source Proxy Decomposition
Proxy Source Cluster Shared W Attribution
Tracing generated output to a single training source remains statistically indeterminate without targeted weight nullspace projection.

Why Deleting Artist Data Retains 88% Style

Read the explanation

In the simulator manifold, a central diffusion probe connects to direct artist training samples and three adjacent proxy clusters. Clicking Delete Artist Training Data severs every direct training link, shrinking their direct weights to zero. Despite total deletion, proxy visual channels expand their attribution, sustaining eighty-eight point four percent style retention in output synthesis.

Super generates helpful tools and automates fact-checking across the internet proactively. If you enjoyed this tool, build your own with Super and share it with a friend.