mapped our assessment against Anthropic's AI Fluency Index — 93% coverage without trying

one of the harder parts of building an assessment tool is knowing whether you're measuring the right things. we've been running aisa.to for a while now (1,800+ assessments completed) and the question kept nagging: how do we validate our framework against what the industry thinks matters?

Anthropic released their AI Fluency Index — a behavioural framework for what AI-fluent people actually do. so we mapped our 11 criteria against their behaviours.

result: 93% coverage. we independently arrived at almost the same picture of what AI skill looks like in practice. the 7% gap is mostly AI governance and organizational policy — stuff that's genuinely hard to assess in a 30-minute conversation.

then we tried the US Department of Labor's AI Literacy Framework (25 sub-competencies). 100% coverage there.

the lesson for other builders: if you're in a space where "what to measure" is unclear, don't start by copying existing frameworks. build from first principles, ship, collect data, THEN compare. you learn way more about which gaps matter vs which are just theoretical.

the other thing nobody tells you — self-reported AI skill correlates with actual performance at r=0.07 to 0.24. basically useless. that's why we went conversational instead of survey-based.

full comparison breakdown: aisa.to/blog/anthropic-ai-fluency-index-comparison