Most people think AI literacy means knowing the tools. Can you write a prompt. Do you know which model is best for what. Have you set up the integrations.
That's worth something. But the tools change every few months, and the person who memorized this season's features is back to square one by spring. Literacy that expires that fast was never literacy.
AI literacy is judgment. It's knowing what to hand to AI and what to keep human. It's looking at a task and seeing, fast, whether a machine should touch it at all, and if it should, where your hand has to stay on it.
AI can make the words. Only a person can make the meaning. A literate person knows which half of that they're holding.
It's a leadership problem first
Here's where this matters more than most people admit.
The conversation about AI literacy almost always points down: train the staff, run a workshop, get everyone using the tools. But the people approving the AI spend need the judgment more than anyone. If the person signing the check can't tell a real capability from a vendor's demo, the firm buys the wrong things. Then it spends a year discovering they don't fit how the work actually runs.
MIT CISR found that only 7% of companies are AI Future-Ready. Most of the other 93% aren't short on tools. They're short on judgment at the level where the money moves.
A board can approve a six-figure AI budget without anyone in the room able to say what they're actually buying or why. That's not a technology gap. It's a literacy gap, and it sits at the most expensive table in the building.
I wrote a longer piece on what this looks like in practice: AI literacy at the board table. The short version is that the people with the least time to learn the tools are the ones whose judgment costs the most when it's wrong.
A firm's literacy is the spread of its people
Individual literacy moves along a ladder. In the Pacemark maturity model, people sit in one of five tiers:
- Non-User — hasn't started.
- Curious — trying things, no real pattern yet.
- Practitioner — uses AI in daily work, gets reliable results.
- Builder — strings tools together, automates, makes things others use.
- Architect — designs how AI fits the whole workflow, teaches the rest.
A firm's literacy is the spread of its people across these tiers. Not the average. The spread.
A company with three Architects and forty Non-Users is not "AI literate" because of the three. It has a few people doing impressive work in isolation and a workforce that can't follow them. The work stays trapped on a few laptops. This is why People & Culture carries the same weight as the technical dimensions in Pacemark: how your people are distributed across these tiers is the real evidence of where the firm stands, and most models skip it.
Here's what that spread looks like in a real firm. We scored a mid-market engineering firm tier by tier: 1% at the top, building tools the rest of the firm uses. 11% building their own workflows. 20% getting reliable results in daily work. 53% still dabbling, mostly using AI as a better search box. 14% who hadn't started. A thin layer of advanced people, a wide curious middle, and most of the firm's real capability resting on a handful of laptops. Read the firm's literacy off that whole curve.
What literate judgment looks like in practice
Judgment shows up in the call you make before you start a task. Hand this to AI; keep that one human.
Hand off the work where the machine is strong and the cost of a miss is low or easy to check:
First-draft synthesis — turning a pile of notes or sources into a structured starting point. Searching across your own documents and past projects to surface what's relevant. The repetitive first pass you'd catch as wrong the moment you saw it.
Keep human the work where meaning, relationship, or accountability is the whole point:
The client relationship and the call that carries real consequences. The judgment you have to be able to stand behind in a room. The work where meaning is the whole point, and a plausible-sounding draft is worse than a blank page.
Then the middle case, where AI drafts and a person owns the result: the proposal that goes out under your name, the analysis a client will act on. The machine gets it most of the way; the last stretch, the part that has to be right, stays yours.
The literate move is rarely "use it" or "don't." It's "use it here, this far, and I check it there." A regional engineering firm I worked with had senior people spending hours digging up old project work to answer new questions. That's a clean hand-off: the machine searches, the person decides what's relevant. The judgment was in drawing that line, not in the tool.
You build literacy as a capability, not a webinar
A one-off training gets you a room of people who watched a demo. Two weeks later, weekly use is back where it started. Literacy is a capability you build over time, the way a team builds any skill that has to stick.
That means practice on real work, not toy exercises. It means people at every tier moving up one step, not the Architects pulling further ahead while everyone else stays put. And it means leaders using the tools in the open first, so the rest of the team has permission to.
This is the work behind the Enterprise Cohort: a team learns together, on their own work, over enough time that the judgment actually forms. If you want the structured version, the AI implementation consulting page lays out how we build it across a firm.
See where your firm stands
You can't build literacy you can't see. Before you train anyone, find out where your people actually sit on the ladder and where the judgment is missing.
The free Signal Scorer gives you a first read in under ten minutes: your Pacemark Index and a radar across the six dimensions, with People & Culture scored alongside the rest. A facilitated Discovery is what maps your whole workforce across the five tiers, but the Scorer shows you where to look first. Start there, then decide what to build.
Literacy is judgment. It's knowing what's yours to keep.
