I spent last week teaching in San Francisco, and the teaching went well, but the thing I keep thinking about happened outside the classroom.
I live in the Midwest, and out here you can go a month without seeing a single advertisement for AI. In San Francisco I could not go a block. It was on billboards, on bus stops, on the sides of buildings, and by the second day I stopped being impressed and started paying attention to what it would feel like to live under it.
Because here is the thing about those ads. If AI has been good to you, they read like progress. If AI is the reason your team got smaller last year, every one of them is a reminder, and you do not get to opt out of your commute. I have heard distrust of AI from students for two years and I always understood it intellectually, but standing in a city where the technology advertises itself on every surface, I finally understood it in my body. The fatigue is not ignorance. It is a rational response to being sold the thing you suspect replaced you.
The fatigue is not ignorance. It is a rational response to being sold the thing you suspect replaced you.
What those ads have in common is not AI. It is dishonesty. Every one of them promises a clean win and never once mentions who paid for it, and that is the same failure I run into at my own desk, just smaller and closer to home, before most people are awake.
I wake up with a hundred ideas. That is barely an exaggeration, and if you are a builder you probably recognize the feeling, because some mornings the list is a gift and most mornings it is noise wearing a gift's clothing. The hard part of my work has never been generating ideas. The hard part is knowing which three deserve to live and which ninety seven are enthusiasm doing an impression of strategy.
The worst person to make that call is me at seven in the morning. I am emotionally attached to all hundred of them, I can talk myself into any of them, and like most solo founders I do not have a disinterested party sitting across the table whose job is to say the uncomfortable thing. Building has never been cheaper or faster, which makes this worse, not better, because the cost of chasing a bad idea used to be a business plan and is now a finished product nobody wanted.
A general assistant tuned to keep you engaged has the same dishonesty as the billboard. It will find something to like about almost any idea in the pile, and agreement feels like validation right up until you ship into silence.
I do not think the answer to either problem is less AI, and I do not think it is more enthusiasm. It is AI that is willing to tell you the truth even when the truth costs it a sale, whether that AI is teaching a room full of people or talking you out of your own bad idea before breakfast. Everything below is what that principle turned into.
Table of contents
- What is the AI Literacy Framework?
- Why does a labor agency care about AI literacy now?
- What are the five foundational content areas?
- What are the seven delivery principles?
- What I built for people using AI at work
- What I built for people building with AI
- Frequently asked questions
- Further reading
What is the AI Literacy Framework?
The U.S. Department of Labor's AI Literacy Framework is voluntary federal guidance that defines what baseline AI skill means for American workers and how to teach it. Issued on February 13, 2026 through Training and Employment Notice 07-25, it names five foundational content areas every worker should cover and seven delivery principles for anyone building the training. Nothing in it creates a new requirement for companies or workers. I treat it as the reference point my own workforce courses get checked against.
Why does a labor agency care about AI literacy now?
The framework did not appear out of nowhere. In April 2025 the White House issued Executive Order 14277 on advancing AI education for American youth. In August 2025 DOL followed with guidance letter TEGL 03-25, which encouraged states to spend Workforce Innovation and Opportunity Act funds on AI skills and sketched a preliminary version of these content areas. The February 2026 framework is the full version of that sketch, published alongside parallel AI education guidance from the Department of Education (DOL news release).
The premise underneath all of it is the part worth sitting with. DOL's position is that every worker will need baseline AI literacy to succeed, regardless of industry or occupation. Offices, manufacturing floors, hospital wings, classrooms. When the federal workforce agency writes that into formal guidance, the question for any working person stops being whether AI skills matter and becomes who is going to teach you, and whether the training is any good. The framework is DOL's answer to the second half, a shared standard that workers, employers, and training providers can all point at.
What are the five foundational content areas?
These are the five things the framework says baseline AI literacy has to cover. Every one of them is skill for a person at a desk, so none of them require a technical background.
| Content area | What it covers |
|---|---|
| Understand AI Principles | How AI actually works, at the mental model level rather than the code level. Probabilistic outputs, training versus inference, capability limits, and the fact that AI can be confidently wrong, which makes verification a core skill. |
| Explore AI Uses | Real workplace applications across drafting, analysis, creative assistance, task automation, and decision support, so workers build judgment about when AI helps and where human input stays essential. |
| Direct AI Effectively | Getting useful results out of a tool. Framing context, structuring prompts, supplying the right input material, and iterating instead of accepting the first answer. |
| Evaluate AI Outputs | Reviewing what AI produces before it moves. Verifying facts, spotting gaps and logical errors, and judging whether an output actually fits the task. |
| Use AI Responsibly | The boundary skills. Protecting sensitive information, following workplace policy, avoiding misuse, and staying accountable for anything you ship with AI's help. |
Notice what the list leaves out. No coding, no model architecture, no math. The framework is explicit that workers need working mental models and judgment rather than technical mastery, and that is a teachable skill set, not a computer science degree.
What are the seven delivery principles?
The second half of the framework is aimed at whoever builds the training, and it reads like a quiet indictment of most corporate AI training that exists today.
| Delivery principle | The idea |
|---|---|
| Enable Experiential Learning | Skill comes from hands-on use on real tasks, with feedback and rising difficulty, rather than from slides about AI in the abstract. |
| Embed Learning in Context | Teach through the worker's actual job, industry, and existing training programs so the content lands as relevant instead of theoretical. |
| Build Complementary Human Skills | Pair AI use with critical thinking, creativity, communication, and domain expertise, since AI amplifies human input rather than replacing judgment. |
| Address Prerequisites to AI Literacy | Check digital literacy, device access, and connectivity first, because AI training fails quietly when the basics are missing. |
| Create Pathways for Continued Learning | Treat foundational literacy as a starting point with visible next steps toward deeper, role-specific skill. |
| Prepare Enabling Roles | Equip the managers, trainers, and counselors who support everyone else, since they shape how AI actually gets adopted. |
| Design for Agility | Build training that refreshes as the tools change, because a fixed AI curriculum goes stale in months. |
What I built for people using AI at work
The framework above is not abstract to me. It is the reference point Working 9 to AI is built against, and the map below shows exactly how.
| Framework area | Where it shows up in the hour |
|---|---|
| Understand AI Principles | You sort your own workload by what AI can hold on its own, what it can help with, and what stays with you, which is capability limits taught through your actual job. |
| Explore AI Uses | The whole session runs on a real task you bring from your week, finished live during the hour. |
| Direct AI Effectively | You leave with prompt templates and guides built to be reused on the next task, so the direction skills survive past the session. |
| Evaluate AI Outputs | A dedicated stretch of the hour goes to catching the places where AI gets your task wrong. |
| Use AI Responsibly | The sorting exercise draws boundary lines, including what should never leave your workplace for a public tool. |
| The delivery principles | One live hour, hands on, on your own work, is experiential and contextual by design, and the seat credits toward the four session Working 9 to AI series for the continued pathway. |
A line about claims, because this matters to me. The framework is voluntary guidance. DOL does not certify courses against it, so no course can accurately claim federal endorsement, mine included. Alignment means the hour was designed to cover the five areas the framework names, delivered the way its principles say training actually sticks, and you are welcome to hold it to exactly that standard.
Working 9 to AI is a four session program for one person at their own desk, with their own tasks and the decisions they make for themselves every day. And its first hour is called Clock In to AI, because the honest place to start is not a lecture about what AI might do someday. It is one hour, live, where you bring a real task from your actual week, finish it with AI, learn to catch the places where AI gets it wrong, and sort your own workload into what belongs with AI and what stays with you. You leave with templates and guides you will actually reuse, and the seat credits toward the full series if you want to keep going.
The first open session is September 17 at 11am Central, online, and it runs the third Thursday of every month after that. A seat is $95. If you have been tired of AI, suspicious of AI, or quietly worried you are behind on AI, this hour was built for you, and I would genuinely love to have you in the room.
Register on Luma ↗ · See the full session details
If you own training decisions for a team, the framework hands you a vendor checklist for free. Ask which of the five content areas any proposed training covers and which delivery principles the format follows. If the answer is a slide deck about AI in the abstract, the framework itself tells you what that is missing. The employer delivery of Clock In to AI runs the same hour for your team on your actual workflows.
What I built for people building with AI
So my morning practice has three rules. The conversation is structured, one question at a time in a fixed order, because a wandering conversation lets me steer toward the answer I want. It is productive, grounded in what the market is already doing rather than in vibes, because pushback means nothing if it is guessing. And it is honest, which means stop is always on the table as a real answer, delivered with a reason I can act on. I wrote about that failure mode of agreeable AI when I first shared the VibeCheck concept, and the months since have only confirmed it.
That practice is now a tool. VibeCheck.exe walks an idea through five questions, one at a time, built on the same ask, explore, reflect structure I teach. It searches what already exists before it forms an opinion, it makes you defend the idea against the hardest argument it can find, and then it gives you a verdict. Sharpen, redirect, or kill. The verdict is free, every time, and kill is a real outcome with a real reason, not a soft no.
If the idea survives, VibeCheck writes the document that usually takes founders a weekend, a build ready PRD you can paste straight into Lovable, Cursor, or Claude Code. That part costs $49 per idea, one time, no subscription, and founding users pay $29 while the launch window is open.
Here is the number I will be watching, and I am putting it in writing so you can hold me to it. The kill rate has to stay above zero. A validation tool that never tells anyone to stop is a yes machine with better fonts, and the whole point of this one is that it is willing to cost me a sale to tell you the truth.
Some of my own hundred ideas have already died in this exact flow, and I mean that as the endorsement it is, because the mornings got lighter when the noise got a fair trial. If you have an idea you have been circling for weeks, give it the check. Worst case, you lose twenty minutes. Best case, you stop building the wrong thing this weekend.
Two rooms, one rule. Whether you are learning to work with AI or deciding whether to build with it, the only version worth trusting is the one willing to tell you no.
Frequently asked questions
Is the AI Literacy Framework mandatory? No. It is voluntary guidance issued through a Training and Employment Notice, and it creates no new regulatory requirements for employers or workers. What it does create is a shared reference for what good AI training covers, which is quickly becoming the standard employers and training providers get measured against.
Who is the framework for? Four audiences by DOL's own structure. Workers building skills proactively, employers training their teams, education and training providers designing programs, and state and local workforce agencies funding all of it.
Does it cover all AI or just generative AI? The framework places its primary emphasis on generative AI, on the reasoning that generative tools are the most widespread and transformative applications and what most people mean when they say AI at work.
Does completing a course make you AI literate under the framework? There is no official credential attached to it. The framework is a map of competencies, so the practical move is to pick training that covers the five content areas through hands-on work rather than lectures, and to expect your skills to need refreshing as the tools change.
What is Clock In to AI? A one hour live session where you bring a real task from your actual week and finish it with AI, learn to catch where it gets things wrong, and sort your workload into what belongs with AI and what stays with you. The first open session is September 17 at 11am Central, $95 a seat, and it credits toward the full Working 9 to AI series.
What is VibeCheck.exe? A guided five question pressure test for startup and product ideas. It researches the market live, pushes back like a sharp cofounder, and gives a free verdict of sharpen, redirect, or kill, then writes a build ready PRD for ideas that survive. It is live at charmthirteen.com/vibecheck.
Why does AI fatigue happen, if AI is supposed to help? Because most AI marketing promises a clean win and never mentions the human cost, and if AI is the reason your team got smaller, every ad is a reminder you cannot opt out of. The fatigue is a rational response to being sold the thing you suspect replaced you, not a knowledge gap.
How much does VibeCheck cost? The verdict is free, always. The PRD unlock is $49 per idea, one time, with a $29 founding rate while the launch window is open. There is no subscription.
Further reading
- TEN 07-25, The U.S. Department of Labor's Artificial Intelligence Literacy Framework (DOL). The primary source, with the complete framework as Attachment I. Read this before any secondhand summary, including mine.
- DOL news release announcing the framework (DOL). The short official version with the five content areas and seven principles named.
- TEGL 03-25, Encouraging the Use of WIOA Funding to Help Youth and Adults Develop AI Skills (DOL). The August 2025 predecessor that sketched the framework and connected it to workforce funding.
- Executive Order 14277, Advancing Artificial Intelligence Education for American Youth (White House). The April 2025 order this guidance chain grew from.
- New DOL Guidance Encourages Employer AI Literacy Training (Ogletree Deakins). An employment law firm's read on what the framework signals for employers.
Changelog
- 2026-08-23, merged the AI Literacy Framework explainer (formerly a separate post, "The Department of Labor's AI Literacy Framework, explained") into this piece, alongside the VibeCheck.exe launch story merged earlier. Framework content covered as issued 2026-02-13 in TEN 07-25, references verified live as of this date. No claims changed.