AI Literacy in 2026: The DOL Says “Learn AI.” Nobody’s Waiting for Permission.

Part 2: The government framework is out. The industry is already moving. Here’s what early movers are doing — and what you can steal from them.

In Part 1, we broke down the five foundational content areas from the U.S. Department of Labor’s AI Literacy Framework — understand AI principles, explore uses, direct AI effectively, evaluate outputs, use responsibly. The short version: the government said every American worker needs baseline AI skills, and then defined what those are.

But the DOL framework didn’t land in a vacuum. It dropped into a world where the EU is already legally requiring AI literacy training, where surveys show people trust AI more than they understand it, and where most workers are figuring this out on their own at 11 PM with a YouTube tutorial and coffee.

This is Part 2. Less framework, more reality.

The Trust Paradox: We’re Confident. We Shouldn’t Be.

Informatica’s CDO Insights 2026 report found that 65% of employees trust the data behind the AI systems they use. Sounds encouraging — until you learn that 75% of data leaders at those same organizations say the workforce needs significantly more AI and data literacy training.

People trust what they don’t fully understand.

Neil Sahota — IBM Master Inventor and UN AI Advisor — calls this the “trust paradox.” It develops gradually. You use AI to rewrite an email. Works great. Summarize meeting notes. Also great. Confidence grows. Then you start using it for client work, compliance decisions, strategic analysis. The confidence you built on easy tasks doesn’t automatically give you the judgment for hard ones.

I’ve seen this play out firsthand. People on my team get frustrated when AI gives confident-sounding answers that are wrong. But when you dig into why, it’s usually one of two things: they didn’t give enough context, or the context itself was wrong. The tool did exactly what it was asked. It was just asked badly.

The DOL framework covers this in “Evaluate AI Outputs” and “Direct AI Effectively.” But the trust paradox adds a critical layer: knowing you should evaluate outputs isn’t enough. You have to actually do it, especially when the output looks right at first glance.

The EU Made This Mandatory. America Is Still on “Voluntary.”

Here’s what didn’t get enough attention: the EU already made AI literacy law.

Article 4 of the EU AI Act took effect February 2, 2025 — a full year before the DOL framework. Any organization building or deploying AI must ensure staff has sufficient AI literacy. Enforcement starts August 2026.

The DOL framework is voluntary guidance.

Meanwhile, the numbers are piling up. The World Economic Forum estimates 44% of workforce skills will be disrupted by 2027. Gartner says 80% of engineering teams need upskilling through 2027. A Bright Horizons/Harris Poll survey found 42% of employees expect their role to change due to AI within the next year — but only 17% use AI frequently today.

PwC reports a 56% wage premium for workers with advanced AI skills.

If you’re reading this, you’re probably ahead of most people. But “ahead” is a moving target.

The Seven Delivery Principles (From Someone Who Didn’t Wait)

The DOL includes seven principles for how AI literacy should be taught. They’re good. Here’s what they look like through the lens of someone learning this on their own.

1. Learn by Doing. The DOL calls it “experiential learning.” I call it “stop watching webinars.”

My background is platform engineering — microservices, infrastructure, and a master’s in data science. When AI tools got good, I didn’t wait for company training. I just started using them.

The biggest discovery: I used to spend days reading product documentation for new tools. I’d understand the config and deployment but still have gaps in how the underlying code worked. With AI, I now get a view into the code itself — the logic, the flow, why things behave a certain way. That gap between “I can configure this” and “I understand how this works” shrank dramatically.

That’s experiential learning. Not from a framework — from needing to get something done.

2. Make It Relevant to Your Job. Generic AI training is like one-size-fits-all shoes. It covers you, but nothing fits. A network engineer and a marketing manager need completely different examples, tools, and risk awareness. If your AI training doesn’t connect to your daily work, you’re not being trained. You’re getting a box checked.

3. Human Skills Matter More Now. PwC shows a 56% premium for AI skills, but WEF data confirms creative thinking, resilience, and leadership remain equally critical. AI amplifies human input — which means it also amplifies the absence of human skills.

When colleagues get frustrated with “generic AI answers,” the root cause is almost always human: they didn’t provide context (communication), didn’t catch errors (domain expertise), or accepted the first response (critical thinking). AI doesn’t replace judgment. It makes the lack of it visible.

4. Not Everyone Starts From the Same Place. Before you learn AI, you need digital skills, a device, and reliable internet. Not everyone has those. I came in with a technical background, so this wasn’t my barrier — but I’ve seen enough colleagues struggle with basic tooling to know this principle matters for real adoption.

5. Don’t Stop at Literacy. This matters most for early movers past the basics. The framework talks about progressing from literacy to proficiency, stackable learning, and even entrepreneurship — building AI-powered solutions, not just using them.

I was running local LLMs with Ollama before my company had an AI adoption plan. Not because I’m some genius — because I was curious and the tools were available. When the company caught up, I had a head start. Build your own path. Stack skills as you go.

6. Train the People Around the Learner. I’ve been the early mover on a team where nobody else was there yet. You learn something useful, you’re excited, and then… you can’t apply it because nobody around you gets why it matters.

The DOL recommends formal train-the-trainer models and peer learning champions. The reality is messier: somebody figures it out, shows results, and others slowly follow. If that’s you — congratulations. You’re the peer learning champion the DOL was talking about. You just don’t have the title yet.

7. Whatever You Build Will Be Outdated Soon. The AI tool you use today might not be the one you use in six months. Invest in transferable skills — clear thinking, good prompting habits, critical evaluation — not in memorizing buttons on a specific tool.

What the Numbers Say (Even If the Framework Doesn’t)

Let me put a few numbers together:

42% of workers say their employer expects them to learn AI on their own. 34% feel unprepared. 79% feel pressure to learn new skills. And the EU has already made AI literacy law while the U.S. keeps it voluntary.

That 42% number sticks with me — not because it’s alarming, but because it describes something I’ve already lived. I started learning AI before my company had a plan. My own time, my own curiosity, a lot of trial and error. That shouldn’t have to be everyone’s path. But right now, for many people, it is.

The DOL framework is one step toward changing that — a common vocabulary for AI literacy that’s less random than “figure it out yourself.”

For the Early Movers

Already using AI regularly? Go deeper. Pick your weakest area from the DOL’s five — for most people that’s evaluating outputs or responsible use — and deliberately practice it. Being fast with AI is easy. Being good takes effort on the boring parts.

Just getting started? Don’t start with a course. Start with a task. One real task from your job. Use AI. Compare. That single experiment teaches more than reading about it.

The “AI person” on your team? Share what you learn. Not “look how smart I am” — more like “I tried this and it saved me an hour.” That’s how AI literacy actually spreads. Not through frameworks. Through someone showing someone else something that works.

No AI training at your company? The DOL framework is free. The EU AI Act guidance is free. Send them to whoever handles training with a note: “we should probably talk about this.”

The framework won’t change the world by itself. But it gives everyone a common starting point. What happens after that is up to us.

Part 2 of a two-part series on the DOL AI Literacy Framework (TEN 07–25). Part 1 covers the five foundational content areas. Full framework at dol.gov.

Sources: Informatica CDO Insights 2026 · Neil Sahota, neilsahota.com · EU AI Act, Article 4 · World Economic Forum · Gartner · Bright Horizons/Harris Poll Education Index · PwC Global AI Jobs Barometer