Someone is selling a $997 AI masterclass right now. While you're reading this sentence, they're posting a countdown timer and a testimonial from someone named "Chad M.from Austin." The course teaches you how to use ChatGPT. ChatGPT, which has a free tier, which you can open in a browser tab right now, which has a help page.
This is the AI grifter economy, and it is enormous. The World Economic Forum estimates AI could affect 40% of jobs globally. That's a real number, and it creates real fear. Fear is a great sales mechanism. The course sellers know this. They built their business on it.
You don't need a $997 course. You don't need a $47 prompt template pack. You need 30 minutes with an actual tool and the confidence to start using it. That's it.
The AI grifter taxonomy: know your enemy
There are five types of people selling AI advice right now. Four of them are wasting your time and money. One is worth listening to.
The Course Seller built a Kajabi landing page and a Stripe account. Their flagship product is a "comprehensive AI mastery program" that teaches you how to ask ChatGPT to write emails. Price point: $497 to $997, with a "fast action bonus" that expires in 48 hours, then resets. They will upsell you into a mastermind group for $297/month where you can ask them questions they answer with ChatGPT.
The LinkedIn Prophet posts every single day about "the future of work." Their content is 80% restatements of things you already read last week and 20% things that aren't true. They have never built anything. They have read everything. They will sell you a newsletter subscription.
The Conference Speaker gives keynotes at industry events about "AI disruption" and "staying ahead of the curve." They use phrases like "we're at an inflection point." They charge $15,000 for 45 minutes. If you ask them what tools they personally use, they get vague.
The Prompt Engineering Guru sells packs of 500 prompts for $47. The prompts were useful for about three weeks in 2023. GPT-4 doesn't need elaborate prompt rituals anymore. The guru keeps selling them anyway.
The Actual Builder uses AI in their real job, posts occasionally about what worked and what didn't, shows their actual work and actual mistakes, and doesn't charge you anything for the information. This person exists on every platform. They're just harder to find because they're not running paid ads.
What AI grifters actually sell (hint: it's fear)
The course exists because you're scared. That's the product. Not knowledge. Fear management.
Every AI grifter funnel follows the same pattern. First comes the alarming statistic, usually from a real study but stripped of all nuance. Then comes the urgent warning: "If you don't learn AI now, you'll be left behind." Then comes the solution: their course, their templates, their mastermind.
Notice what's missing. Any evidence that the thing they're selling actually helps.
The Goldman Sachs report on AI and jobs that grifters love to cite? It also says AI will create new roles and that most workers will see task automation, not job elimination. They leave that part out. You'd have to look it up yourself.
The tell is always in what they don't show you. They show screenshots of ChatGPT conversations as "results." They don't show you their client list, their actual revenue from using AI (not from selling AI education), or what their work looked like before and after. Because the work is selling AI education. The AI is incidental.
Red flags: five ways to spot one before you hand them money
You'll get faster at this. Once you see the pattern, you can't unsee it.
They sell fear first. The hook is always some version of "you're about to be left behind." Legitimate educators don't need to scare you into buying. The fear is the product; the course is just the receipt.
Their "proof" is ChatGPT screenshots. Real results look like: before I used AI for X task it took me 4 hours, now it takes 45 minutes. Fake results look like: here's a screenshot of ChatGPT writing a pretty good LinkedIn post. One of these is evidence. The other is a demo of a free tool.
They teach you features, not judgment. Any course that promises to teach you "100 AI prompts" is teaching you the wrong thing. The prompts go stale. The judgment about when and how to use AI doesn't. If the course is a list of things to copy and paste, skip it.
There's always an upsell. The $97 course leads to the $497 course leads to the $297/month mastermind. The business model requires you to keep buying. A resource that actually makes you competent doesn't need you to keep paying for it.
They've never built anything with AI. Ask them. Seriously. "What have you shipped using AI tools?" If the answer is "I've created courses and content about AI," that's your answer. They study the tools. They don't use them to do actual work.
This came from a book.
Don't Replace Me
200+ pages. 24 chapters. The honest version of what AI means for your career, written by someone who actually builds this stuff.
Get the Book →What actually works (and costs nothing)
Here's the uncomfortable truth for anyone who just bought a course: the best way to learn AI tools is to open them and use them.
The official OpenAI documentation is free and well-written. Anthropic's docs for Claude are the same. YouTube has tutorials from actual practitioners who are not selling you anything beyond watch time. And the tools themselves have built-in help, examples, and suggested prompts.
The learning curve is roughly: 30 minutes of actual use gets you functional. Two weeks of daily use gets you good. There is no curriculum between those two milestones. You just have to do the thing.
What you actually need to learn is covered in the AI skills guide for non-technical people on this site. No coding, no courses, no credential. The short version: pick one tool, pick one annoying task in your current job, and spend a week trying to make that task faster. That's the whole curriculum.
Dee covers the logic behind this directly in Don't Replace Me, specifically the rule about forgetting "learning AI" as a goal. The point isn't to become an AI expert. The point is to stop wasting time on the parts of your job that a good tool can handle in 90 seconds.
The prompt engineering lie deserves its own section
Prompt engineering was briefly real. For about 18 months, knowing how to structure requests to early GPT models produced meaningfully better outputs. You could get genuinely better results with a well-crafted prompt versus a lazy one.
That gap has mostly closed. Modern language models are much better at understanding intent from normal, plain-English requests. The elaborate ritual of "act as a senior copywriter with 20 years of experience and respond in a professional but approachable tone" is mostly theater now. You can just say what you want.
The people still selling prompt engineering courses know this. They're selling you a skill that was marginally useful in 2022 and is largely unnecessary now. The market has moved. Their course hasn't.
The MIT Technology Review covered this shift in some depth. The short version: as models improve, prompt complexity requirements decrease. You don't need a guru. You need to describe your actual problem in normal language.
How to actually get good at using AI at work
You don't need a course. You need a starting point. Here's one.
Think of the last task at work that made you feel like you were doing something a moderately smart intern could handle. Writing a summary of a long document. Drafting a routine email. Building a first draft of a presentation you've given 12 times. Turning a mess of notes into a formatted report.
Pick one of those. Open ChatGPT or Claude. Describe what you need. See what comes out. Adjust. Try again.
That's it. That's the whole process. The practical guide to using AI at work walks through the most common use cases if you want something more structured, but the real version is: start with the thing you hate most, use the tool, see what happens.
You will get bad outputs sometimes. That's part of using the tool, not evidence that you need a course to use it correctly. Everyone gets bad outputs. The practitioners who are good at this get bad outputs and know how to iterate from them. That skill comes from repetition, not from paying $997.
And if you want specific prompts for real office work, the ChatGPT prompts guide for work has actual examples without the mastermind price tag.
The real cost of the grifter economy
Here's what bothers me most about this. It's not the money, though $997 is a lot of money to spend on a course teaching you a free tool.
It's that the AI panic machine makes competent people feel helpless. You see the countdown timer and the alarming statistic and the "fast action bonus" and your brain reads: this is complicated, I need an expert, I'm already behind. None of that is true. The complexity is manufactured. The expertise is mostly recycled documentation. The deadline is fake.
You are not behind. The tools are genuinely not that hard to use. The people who will struggle with AI over the next five years are not the ones who skipped a masterclass. They're the ones who let the fear run so deep that they never bothered to open the tool.
One open browser tab beats a $997 course every single time.
Frequently asked questions
Are AI courses worth paying for?
Almost never. The free resources available directly from OpenAI, Anthropic, and YouTube practitioners are better than most paid courses and cost nothing. If a course is charging you to learn ChatGPT, which has its own free tutorials, the course is the product, not the education. The only paid AI education worth considering is narrow, technical, and specific to something you're already actively doing.
What is prompt engineering and do I need to learn it?
Prompt engineering was the practice of carefully crafting requests to get better outputs from early AI models. It mattered more when models were less capable. Modern models like GPT-4 and Claude understand plain-English requests well enough that elaborate prompting rituals are largely unnecessary. You don't need a prompt engineering course. You need to describe what you want clearly.
How do I spot an AI grifter before I give them money?
The fastest tells: they sell fear before selling anything else, their "proof" is screenshots of ChatGPT outputs rather than real work results, there's always an upsell to a more expensive product, and when you ask what they've actually built with AI (not sold about AI), they get vague. Real practitioners show their work. Grifters show testimonials from people who liked the course.
What's the best free way to learn AI tools for work?
Open the tool and use it for one real task from your actual job. ChatGPT and Claude both have free tiers, both have their own documentation and suggested prompts, and both produce useful output within minutes of starting. See how to use AI at work for a practical starting point, or the ChatGPT guide for office work for specific prompts.
Is prompt engineering dead?
Largely, yes, for everyday use. MIT Technology Review reported on this shift as models have become significantly better at parsing intent from normal requests. Highly technical applications still benefit from precise prompting, but the kind of prompt engineering being sold in most courses is no longer necessary for typical office tasks.
What should I actually spend money on to stay relevant with AI?
Nothing, to start. Use the free tools for 30 days before spending anything. If after that you find a specific gap in a specific skill, target that gap precisely. A narrow technical course on a specific tool you're already using is different from a broad "AI mastery" program. The skills guide for non-technical people outlines what actually matters without the upsell.