Hit 1,800 assessments on aisa.to this week. Been building this AI skills assessment tool for a while now and the dataset is finally big enough to see patterns I wasn't expecting.
The one that caught me off guard: product managers score 59.2/100 on average. Engineers score 53.7. I assumed technical roles would dominate but PMs are pulling ahead — mostly because they're better at knowing WHEN to use AI and evaluating whether the output is actually good.
Other stuff from the data: average score across everyone is 48/100. 67% score below Proficient. People overestimate their own skills by ~19 points. Biggest within-team gap we've seen: 82 points between best and worst on the same team.
The overestimation gap keeps surprising me. Almost everyone thinks they're better with AI than they actually are. Not because they're bad — because "I use ChatGPT daily" feels like skill but it's really just usage.
Full data breakdown: https://aisa.to/ai-fluency-index
Anyone else building in the AI skills/assessment space? Curious what patterns you're seeing.
The 19-point self-assessment gap is probably more commercially interesting than the PM vs engineer result. If usage is such a weak proxy for actual capability, I’d be curious whether teams know what to do differently once they see their assessment results.