The AIFluency MapHow good are you at AI?
AI is not a self‑driving car.
What AI fluency is
AI fluency is the ability to work with AI effectively, safely, and adaptably, across tools, across contexts, and under ambiguity. It also includes AI judgement: knowing when, and how, not to use AI.According to aisa.to
What AI is and is not. How it fails.
Asking well. Checking what comes back.
Working with it every day, in your role, at your stakes.
aisa.to/ai-fluency. Literacy is reading. Fluency is writing and speaking.
What makes a good driver
Goals for Driver Education (GDE) matrix, from the EU GADGET project: Hatakka, Keskinen, Gregersen, Glad & Hernetkoski (2002), Transportation Research Part F, 5(3), 201–215. AISA rubric: 5 dimensions, 11 criteria, aisa.to/resources/the-aisa-rubric.
Five dimensions, eleven skills
- Prompt design
- Iterative dialogue
- Context & memory management
- Output evaluation
- Limitation awareness
- AI fundamentals
- Tool landscape
- Workflow integration
- Task decomposition
- Domain application
- AI safety & responsibility
Averages across 2,752 AISA assessments.
How AISA works
The five dimensions you just saw. Eleven criteria. One fixed, published rubric.
A twenty-minute conversation with Aisa. Not a quiz, not a survey. A second AI scores every answer as you talk; a third calibrates the result.
four drivers
One request, “help me write a social media post promoting our event”, lands with four people scoring 34, 55, 70 and 85. You say how you would do it first. Then we climb.
Open the game →Where do you stand?
2,752 AISA assessments. Personas placed at their average score.
Ten AI personas
A persona is assigned by the shape of your profile across the five dimensions. The ranges are typical scores, not hard bounds.
The map: from here to fluent
Your 5 tools to master AI (and any other skill).
Role based averages
Average dimension score (0–100) by role, roles with 50+ assessments, across 2,752 AISA assessments. Role taken from the candidate’s own job title. Darker is lower.
Role AI fluency archetypes *experimental
The minimum profile a role needs, not what people score today.
Our working view of where each role needs to be, drawn from how the work gets used and who it reaches. To be tested against the role averages as the data grows.
Five points
of insight.
Every person answers one research question before the assessment begins. One question at a time, retired at 100 answers, across 2,752 assessments so far.
AHas AI changed the way you think about your job?
Not at all7%A little15%Meaningfully36%Fundamentally31%I'm still figuring that out11%
Two thirds say AI has changed how they think about their job meaningfully or fundamentally.
One answer per person, chosen before the assessment. Options shown in the order they were asked.
BHow do you typically learn about new AI capabilities?
Social media37%Colleagues13%Newsletters6%I experiment myself37%I don't actively follow it7%
Social media and self-experiment tie at 37 each. Newsletters barely register.
One answer per person, chosen before the assessment. Options shown in the order they were asked.
CDo you think AI skills should be a factor in hiring decisions?
Definitely yes60%Probably yes22%Neutral12%Probably not3%Definitely not3%
82 in 100 say AI skills should count in hiring. Six say no.
One answer per person, chosen before the assessment. Options shown in the order they were asked.
DWhat's the AI skill you most want to develop next?
Prompt engineering22%Building AI agents or automations50%Evaluating AI output critically11%Understanding AI architecture14%Knowing when NOT to use AI3%
Half want to build agents or automations. Knowing when not to use AI: one person so far.
One answer per person, chosen before the assessment. Options shown in the order they were asked.
EWhat's your AI hot take?
AI will fundamentally transform most jobs within 5 years47%AI is genuinely useful but won't replace people34%AI is overhyped — the reality doesn't match the buzz2%I'm more concerned about AI's risks than excited about its potential7%I don't have a strong view yet — still forming my opinion10%
Nearly half expect AI to transform most jobs within five years. Two in 100 call it overhyped.
One answer per person, chosen before the assessment. Options shown in the order they were asked.
An example from my workflow
A skill I open at the start of the day, /desk. Iris is a persona with her own charter and her own memory.
- She reads six desk files, and nothing else: to-do, rolodex, waiting-on, reminders, notebook, whiteboard. Never the code.
- She sequences the to-dos so they get done. Sized to 10–15 minutes, the easiest start first, ordered for the most satisfying run of closes: four to six a block.
- She is built in concentric circles. Her charter, who she is and how she works, comes before her tasks; her memory wraps around both. The character is set first, the work happens inside it.
- She learns on her own. Every session she records what she did well and what she did not, and writes it to her memory, so the next morning starts further along.
The 70 move from the game: context built once, reused every morning, and it improves on its own.
AI is not a self‑driving car.
Q&A
Complete the assessment to see your report, and start the AI Fluency Voyage game.
