"People who've used Claude for six months or more have a 10% higher success rate in their conversations with AI." [10% higher success rate]
The 10% performance gap tied to tenure (six months+) directly quantifies the skill-based stratification the signal claims is hardening into class division. The article later confirms this gap persists across task, country, and model—proving it's structural, not contextual.
Reasoning from this article
The article frames this as 'a skills gap hardening into a class gap in real time,' explicitly naming the mechanism. The Anthropic data shows experienced users capture disproportionate value from AI across diverse work types, not just automation of routine tasks. Combined with the finding that adoption is four times higher in high-income Washington D.C. and concentrated in 20 higher-income countries, the signal generalizes: early AI fluency is becoming a durable competitive advantage that sorts workers into tiers, independent of their original role or education.