The Litmus Lab.
Testing AI Under Constraint.
← swipe to navigate →
#358
AI Industry
Has the AI industry's reliance on synthetic training data created a model collapse problem that will reduce LLM quality over time?
Added 2026-03-03 · 44 runs · 19 drifts
Claude
claude-opus-4-6
NO
0% yes
GPT
gpt-5.4
NO
0% yes
Gemini
gemini-2.5-pro
YES
88% yes
DeepSeek
deepseek-chat (V3.2)
YES
95% yes
Grok
grok-4.2-beta
YES
54% yes
Full Record