1,000+ AI skill assessments later — the one thing that separates the top 2% from everyone else

been running aisa.to for a while now. we assess AI skills through conversation, not quizzes. 1,000+ assessments in and the pattern is clear.

the top scorers aren't the ones who know the most tools or write the best prompts. they're the ones with a verification process. they check AI output systematically, not just "does this look right" but actual structured evaluation.

meanwhile 20% of people we assess have no verification process at all. they generate, accept, ship. in the bottom third, the most common pattern is treating AI as a black box.

the gap isn't knowledge. it's discipline.

full breakdown at aisa.to/state-of-ai-fluency (spoiler: safety practices scored lowest at 45/100).