A lot of AI training feels like a tour of a fancy new kitchen.
Here’s the oven.
Here’s the blender.
Here’s the drawer full of tiny tools nobody understands.
Good luck making dinner.
That may help people find the buttons. It doesn’t help them cook. And that is the problem with a lot of AI literacy training right now. It teaches people how to open the tool, type a prompt, and act impressed when a tidy answer appears twelve seconds later.
Useful? Sure. Enough? Not even close.
AI literacy is not memorizing five prompts. It is not watching someone demo ChatGPT for 45 minutes. And it is definitely not learning that if you type “act as an expert,” the machine suddenly becomes one.
That last one would be nice, though.
Real AI literacy is knowing what AI is good at, where it gets shaky, how to guide it, how to check it, and when to leave it sitting quietly in the corner. Because AI makes it very easy to create work that looks finished before anyone has thought deeply about whether it is good.
A course outline.
A policy summary
The Real AI Skill Is Judgment
A set of quiz questions.
A manager email.
A slide deck with 44 bullet points and the word “leverage” nine times.
AI can produce all of that in seconds.
So the better question is not:
Can people use AI?
Most people can figure out the basics.
The better question is:
Can they judge what comes back?
That’s the gap.
The Real AI Skill Is Judgment
The U.S. Department of Labor recently released an AI Literacy Framework. The part I like most is that it does not treat AI literacy like basic software training.
It points to a better set of skills: understand what AI can and cannot do, explore where it fits in real work, direct it with clear context, evaluate what it produces, and use it safely and responsibly.
That is a much better map.
Because people do not usually fail with AI because they forgot which button opens the sidebar. They fail because they trust the confident answer. They fail because they ask vague questions and get vague sludge. They fail because they paste sensitive data into the wrong place.
And they fail because the output sounds polished, so their brain takes a nap.
That last one is sneaky.
Polish creates trust.
Even when it should not.
We have all seen it. The AI answer arrives neat, fast, and strangely confident. It has headings. It has bullets. It has the rhythm of someone who owns a blazer and says “circle back” without shame.
And because it looks competent, we treat it as competent.
That is where training has to get sharper.
Don’t Teach Prompts First
I am not anti-prompt.
A good prompt is useful.
So is a good wrench.
But if I hand you a wrench, you are not suddenly a plumber. You are a person holding a wrench near a future insurance claim.
Same with AI.
Prompting matters, but it is not the main event. The main event is helping people understand the work well enough to know what good looks like.
That means AI literacy training should focus less on magic phrases and more on real decisions:
- Should AI be used for this task?
- What context does it need?
- What would a good answer include?
- What could be wrong, missing, biased, outdated, or risky?
- Who needs to review this before it goes anywhere important?
- What part still requires human judgment?
That is not tool training.
That is work training.
And L&D should be leading it.
Try the Trust, Tweak, or Toss Test
Here is a simple way to make AI literacy more practical.
Teach people to sort AI output into three buckets:
- Trust it.
- Tweak it.
- Toss it.
Trust it when the output is accurate, useful, safe, and matched to the task.
Tweak it when the bones are good, but it needs better context, clearer language, stronger evidence, or a more human point of view.
Toss it when it is wrong, risky, generic, biased, made up, or pretending to know something it does not know.
That tiny habit does a lot of work.
It slows people down. It gives them a decision path. It reminds them that AI output is not a finished product.
It is raw material.
Sometimes useful raw material.
But still raw.
A Better Way to Design AI Literacy Training
Instead of building a generic “Intro to AI” course, start with the work people actually do.
Pick one task your audience touches every week. Not glamorous work. Real work.
Summarizing meeting notes. Drafting customer responses. Creating first drafts of training materials. Reviewing data. Preparing manager talking points. Writing status updates that do not make people want to walk into the sea.
Then build the training around that moment.
Show the messy version. Add AI as a helper, not the hero. Show a good output. Show a bad output. Ask people to decide: trust, tweak, or toss?
This is where the learning happens.
Not when the AI answers.
When the human evaluates.
That is the muscle we need to build.
The Friday Find
This week’s find is the Department of Labor’s AI Literacy Framework.
I do not usually recommend government PDFs as light weekend reading. I have boundaries.
But this one is worth a look because it gives L&D teams a cleaner way to think about AI training.
The big takeaway:
AI literacy is not about creating power users of a tool.
It is about creating better thinkers, reviewers, and decision-makers in AI-shaped work.
That is a very different design challenge.
And a much more interesting one.
Worth Your Attention
Start Here: The AI Literacy Framework
This is the cleanest map I’ve found for what AI literacy should actually include. Not just prompts and tool tips, but the bigger skills people need: knowing what AI can do, where it fits, how to guide it, how to check it, and how to use it responsibly.
Short on time? Here’s the infographic version.
See the AI literacy framework →
AI Should Make Learning Design Better, Not Just Faster
Dr. Philippa Hardman nails the bigger opportunity: AI is not a shortcut around learning design. It is a chance to get more serious about it.
Use AI to design better, not just faster →
How to Use AI Without Handing Over Your Judgment
A smart look at where AI can help instructional designers with research, ideas, critique, and production while keeping the human in charge of the decisions that matter.
The Bottom Line
AI literacy is not about turning people into prompt wizards.
It is about helping them become better thinkers, reviewers, and decision-makers in AI-shaped work.
The tool can draft, summarize, and suggest, and it sounds very confident while doing so. But the human still has to decide whether the answer is useful, accurate, safe, and worth putting their name on.
That is the skill to build.
Not just:
“How do I use AI?”
But:
“Should I trust it, tweak it, or toss it?”
That is where AI literacy gets real.
🎵 Today I’m listening to Arcy Drive
📍Where I’ll be next
ATD Core4, Sept 29-Oct 1(Indianapolis, IN)
Found this useful? Forward it to someone else who’d also get something out of it.