1,306 AI assessments in — the skill gap nobody talks about
crossed 1,300 assessments on aisa.to. some numbers that keep showing up:
average score: 52/100. 62% of people below Proficient. lowest dimension: Safety at 45/100.
that last one surprises people. everyone focuses on prompting or tool selection, almost nobody thinks about when NOT to use AI, or how to verify whether the output is safe to ship.
the highest scorers aren't the most technically skilled. they're the ones who systematically check output before using it. boring process stuff, but it's the actual differentiator.
building out the AI coach feature now — trying to turn assessment data into real behaviour change, not just a number. that's the hard part.
Safety at 45 tracks. But if your best scorers are just the people who check their work, that's a habit and not knowledge.
The "Safety" dimension scoring lowest (45/100) is such an underrated insight. Most AI tutorials teach you how to get better outputs — very few teach you how to know when the output is wrong or dangerous to ship.
I've noticed the same pattern building an AI content tool for small business owners. The users who get the most value aren't the ones who craft the best prompts — they're the ones who developed a habit of reviewing and editing the output before posting. That verification layer makes all the difference.
Curious: in your data, did you see any correlation between industry/use case and Safety scores? I'd imagine people using AI for customer-facing copy (marketing, support) might be more aware of the risks than those using it internally.