Unverified50% confidenceFactExact time
The study additionally introduced six deliberately weakened models with clearly inferior capabilities as positive controls.
1
Sources
50%
Confidence
Long-term
Relevance
9/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe user stated that the capabilities of Deep Research mode appear to have been degraded by approximately 10x over the past 6 to 7 months.69% similarVerifiedA 2023 joint study by Oxford University, Cambridge University, and other institutions elaborated on the model collapse effect: when a model is iteratively trained on its own generated data, the statistical variance of outputs continuously decreases, rare long-tail knowledge disappears first, and the model ultimately converges to a cultural mean state.68% similarUnverifiedThe same evidence can improve the performance of weaker models while degrading the performance of stronger models.67% similarUnverifiedAnthropic's research shows that even mild alignment training can cause a 2–8 percentage point decline in model performance on certain reasoning-intensive tasks.67% similarUnverified研究发现当约束条件从1个叠加至5个时，几乎所有模型的遵循率都出现断崖式下降，大模型更明显67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/924829API
curl https://kongchang.com/api/v1/knowledge/claims/924829MCP
get_claim(id=924829)