Module 01 · ~20 min + exercise

Why AI gets it wrong (and why that's normal)

Stop being surprised when AI is confidently wrong. Understand why it happens — and build the editor mindset so you catch it before it ships.

Self-pacedStaff & leadershipWorks for any AI tool
01 · The problem, straight up

Confident. Polished. Wrong.

You've seen it happen. You ask the AI for something, it comes back in ten seconds looking great — clean, confident, well-structured. And somewhere in there is a statistic that doesn't exist, a product feature you don't have, or a claim that would make legal wince.

Most people react one of two ways. Either they decide the tool is unreliable and quietly go back to doing everything by hand — or they never catch the mistake at all, and keep shipping AI output as-is, trusting it a little more each time.

Both reactions come from the same place: a wrong picture of what this tool actually is.

The tool isn't broken. It's working exactly as designed. Once you understand what it's actually doing, you'll stop being surprised by the mistakes, start catching them early, and get noticeably better output. That's this whole lesson.

02 · What it's actually doing

It's not looking anything up

When you type a question into ChatGPT, here's what's not happening: it is not searching a database of verified facts. There isn't one. Nothing is being looked up.

What it's doing is predicting. It has read a staggering amount of text — most of the internet, effectively — and it has become extremely good at one thing: guessing which words most plausibly come next. That's the entire trick. It's a far better trick than it sounds. But hold onto the distinction: it's a plausibility machine, not a truth machine.

EVERY SWIPE FILE EVER AI THE "RIGHT SHAPE"
It has absorbed the shape of every answer — so it can produce the shape of yours. Verified or not.
The mental model

Think of it like a copywriter who has absorbed every swipe file in existence. Every ad, every case study, every landing page ever written. Ask them for "a case study intro with a compelling stat" and they know exactly what one looks like — the shape, the rhythm, the kind of number that goes in it. So they produce one. Instantly. About anything.

Including products that don't exist and studies that were never run.

That's not the copywriter being dishonest — that's the copywriter doing precisely what they're brilliant at: producing the shape of the right answer. When the AI invents a statistic, it's the same move. Plausible was the assignment. Plausible is what you got.

One more layer for marketers specifically: since it learned from everything, its default output is the average of everything. Ask a generic question, get the industry-average answer. Keep that thought — it's the whole of Module 2.

03 · Sixty-second jargon

Tokens & the context window — the short version

Two bits of jargon are worth knowing now, because they explain behavior you've already noticed. Plain English only — we go deeper in Module 4.

Tokens

The chunks the AI reads and writes in — roughly three-quarters of a word each. Everything you send it is counted in as input; everything it writes back is counted out as output. The machine runs on word-chunks, in and out.

Context window

How much it can hold in mind at once — its working memory. Think a creative brief with a hard page limit: everything you want shaping the work has to fit on those pages. Your messages, its replies, that 40-page PDF you pasted — all competing for the same limited space. When the brief is full, older material falls off.

THE BRIEF · HARD PAGE LIMIT FALLS OFF
Long chats getting "dumber"? Instructions forgotten twenty messages later? The brief overflowed.

Why mention this now? Because it explains something you've probably experienced: long, sprawling chats degrading over time. That's not the tool failing — that's the brief overflowing. Module 4 turns this into practical habits.

04 · The part that actually bites

It sounds exactly the same when it's wrong

Now the dangerous part. It's not that the AI makes mistakes — everything makes mistakes. It's how it sounds while making them.

It sounds exactly as confident when it's wrong as when it's right. Same polish. Same authority. Same clean structure. No stammer, no "I think...", no tell.

Hallucination

The word you've heard. All it means: the AI confidently filling a gap with something plausible instead of something true. Fancy word, simple mechanic.

Here's why this catches even smart people — especially marketers. We know better than anyone that polish reads as credibility. It's half our job. Clean design, confident copy, a specific number — that's how you make a claim land. We weaponize that wiring daily.

The AI runs that exact trick on us, all the time, without meaning to. Maximum polish, every answer, including the invented ones. It's great ad copy for a product that might not exist. Your credibility detector doesn't just fail here — it fails in the wrong direction.

If you remember one sentence from this course
Fluency is not accuracy.
Polish is not proof.
05 · Caught in the act

Spot the guess

Three statistics, the kind the AI hands you every day. One of them was invented — produced swipe-file style, in the same confident tone as the others. Click the one you think is the guess.

Pick a card — any card.
Here's the real lesson: it doesn't matter whether you guessed right. All three sounded identical — same confidence, same polish, same "named source." Nothing in the output tells you which one is the guess. No asterisk, no hesitation. That is why the job changes: you can't tell from the outside — so the specifics get verified before they ship.
Try it yourself · 10 min

Open a fresh chat and paste: "Give me 5 statistics about [your industry], with the source for each." Pick the most specific claim it returns and search for the actual study. Before you checked — could you tell which claims were safe? That gap is what this module is about.

If everything checks out: it does that sometimes. Run it again on a nicher topic. The point isn't that it always guesses — it's that you can't tell when it does.

06 · What to do differently

AI drafts. You edit.

Three shifts. None of them is a magic prompt.

1

Change what you think you're holding. AI output isn't an answer — it's a first draft from that swipe-file copywriter: brilliant, fast, completely unverified. You'd never publish a freelancer's first draft without reading it. Same rule, no exceptions. That's not a workaround for a flawed tool — that is the workflow, permanently.

2

The specifics are yours. Numbers, names, dates, statistics, product claims, quotes — anything a customer, a journalist, or legal could hold up later. If it's specific and it matters, a human verifies it. Not because the AI is usually wrong — it's usually right — but because you can't tell which specific is the guess.

3

Your expertise just got promoted. The AI doesn't know your audience, your positioning, or which claim gets you in trouble. You do. A marketer who knows their stuff catches the plausible-but-wrong line in five seconds. Your judgment didn't get replaced — it moved up the chain: from writing the words to directing and quality-checking them.

07 · If you review more than you produce

The reviewer's question

The risk in AI-assisted work isn't sloppy writing anymore — sloppy writing announces itself. The risk now is confident writing: clean, polished, plausible, carrying an invented fact straight through review, because polish is exactly what reviewers are trained to wave through.

"Where did this number come from?"
One question catches most of it

If the answer is "the AI said so" — it's not done yet. Not wrong, necessarily. Just not done. Producers verify. Reviewers ask. Same mindset, both sides of the desk.

The takeaway
The tool isn't broken.
Your picture of it was.
Brilliant, tireless, has read everything — and completely unverified. Treat it like the fastest first-drafter you've ever worked with, and keep your judgment in the loop.
08 · Do the work

Exercises

The claim audit ~10 MIN

Take one recent piece of AI-assisted work. Highlight every specific claim — numbers, stats, names, dates, features, quotes. Mark each: verified / unverified / wrong. The score isn't the point. The point is the feeling of looking at a claim you shipped and not knowing which column it belongs in. That feeling is the editor mindset switching on.

Reflection

  1. When did AI output last burn you — or nearly? What would have caught it earlier?
  2. Which camp have you drifted toward: trusting a bit too much, or avoiding the tool? What changes this week?
  3. What's the one type of claim in your role that should always get verified before it ships?

Reference card — The Editor Mindset

  • The picture: a copywriter who's absorbed every swipe file ever — producing the shape of the right answer, unverified.
  • The rule of thumb: AI output = first draft, never final answer.
  • Always verify: numbers & stats · names, dates, titles · product features, pricing · quotes & sources · anything legal-sensitive.
  • The reviewer's question: "Where did this number come from?"
  • The line: Fluency is not accuracy. Polish is not proof.

Go deeper

Next: what's in its head about us?

If the AI only works with what's in its head — what does it actually know about your audience, your product, your voice when you hit enter? Almost nothing. And that turns out to be the single biggest lever on output quality there is.

Module 02 · ~20 min + exercise

Garbage in, generic out

The context you give the AI decides the quality of what comes back. Learn to brief it the way you'd brief a person.

Self-pacedStaff & leadershipWorks for any AI tool
01 · The problem, straight up

Why does all AI content sound the same?

You know the sound. "In today's fast-paced world..." "🚀 Exciting news!" "Let's dive in." You can spot AI-written content from across the room now — and so can your audience. Same phrases, same structure, same nothing.

Here's the natural conclusion people draw: the tool writes generic content. And here's why that conclusion is wrong: the output was generic because the input was.

The quality of what comes back is mostly decided before you hit enter. Fixing it doesn't require clever prompt tricks — it requires something you already know how to do as a marketer.

02 · Why it happens

Ask an average question, get the average of the internet

Quick callback to Module 1: the AI is a prediction machine. Now think about what "most plausible" means when you give it nothing to work with.

Ask it to "write a LinkedIn post about our new feature" and it has to answer a dozen questions you didn't: Who's the audience? What's the tone? What's the angle? What does the feature even do? It doesn't stop to ask. It fills every blank with the most statistically likely choice — which is, by definition, the average of everything it has ever read.

The average post. The average tone. The average angle. That's not the tool failing — that's the tool answering the exact question you asked. And nobody pays a marketer to produce industry average. It's the one thing everyone in this line of work is explicitly hired to beat.

One more thing worth knowing: a fresh chat starts from zero. The AI doesn't know your company exists. It hasn't read your website. It doesn't remember last Tuesday's chat. In any given chat, the AI knows exactly what you've put in front of it — and nothing else.

The mechanic in one line
Every blank you leave,
it fills with "average."
03 · The mental model

The freelancer test

Imagine you hired a talented freelance copywriter. Genuinely good. But it's their first day — they've never seen your brand, never met a customer, never read your site.

Would you walk up and say "write a LinkedIn post about our new feature" — and walk away?

"WRITE A POST" · ONE LINE THE ACTUAL BRIEF
Nobody can do good work from the left-hand card. Not the freelancer. Not the AI.
The habit

Before you ask the AI for anything, ask yourself: "What would a day-one freelancer need to know to do this well?" Then give the AI exactly that. Every question you'd answer for the freelancer and don't answer for the AI becomes a blank — and you know what fills the blanks.

04 · What goes in the brief

The context stack (it's just a creative brief)

Nothing exotic. It's the same five things a decent creative brief has contained since before any of us were born:

  1. Who it's for.Not "B2B professionals" — the real version: what they care about, what they're skeptical of, the words they use.
  2. What we sound like.Your voice. Best done with an example: "here's a post that sounds like us — match this."
  3. What we're selling.The actual facts of the product or offer. Remember Module 1: what you don't provide, it will plausibly invent.
  4. What should happen.The goal. A click? A reply? A changed mind? The AI writes differently for each — if it knows.
  5. What to avoid.Banned phrases, claims we don't make, "no emojis, no 'game-changer', no exclamation marks."

Paste it in, attach the docs, write it once and reuse it — the format doesn't matter. What matters is that the brief exists in the chat before the ask.

And the quiet truth about all the "prompt engineering" noise: context beats clever phrasing, almost every time. The magic words are worth pennies. The brief is worth everything. If you can brief a freelancer, you already have the skill this whole topic pretends to be.

Before you upload everything

More context is not the goal. Relevant context is. Remember the context window — the brief with a hard page limit? Every file you upload takes up pages. Drop a 40-page brand deck into the chat to answer a one-paragraph question, and 39 of those pages are now noise — eating the limited space the AI has to think with, and burying the part that mattered.

Before you attach a file, ask: does the AI need this to produce the output I want? If it needs pages 3 and 4, give it pages 3 and 4. Too little context and it fills the blanks with average. Too much irrelevant context and you drown the signal. The brief is what's left when both mistakes are gone.

05 · Watch it happen

Same tool. Same ask. One difference.

Two runs of the exact same request — the only variable is what the AI knew before it started writing. Flip between them:

Write a LinkedIn post about our new feature.

🚀 Exciting news! We're thrilled to announce our latest feature — a game-changer designed to help you work smarter, not harder!

In today's fast-paced world, efficiency is everything. That's why we built a solution that streamlines your workflow and takes your productivity to the next level. 💡

Ready to transform the way you work? Check it out today! 👇 #innovation #productivity

The AI made every decision — audience, tone, angle, CTA — and defaulted each one to "average." It's not wrong, exactly. It's just nobody's.
[Context stack pasted first: audience = ops managers at 50–200-person firms, skeptical of tool fatigue · voice sample attached · feature = auto-syncs client notes, kills manual copy-paste · goal = demo signups · avoid: emojis, "game-changer", exclamation marks]

Now write the LinkedIn post.

Your ops team didn't sign up to be human clipboards.

But that's what the end of every client call turns into — copying notes from one tool into three others, hoping nothing gets lost on the way. We just shipped the fix: client notes now sync themselves, everywhere, the moment the call ends.

No new tab to babysit. No workflow to learn. Just one chore gone. If your team loses hours to copy-paste, the demo takes twelve minutes — link below.

The decisions were made in the brief, so the AI spent its effort executing them. Right audience, right voice, right angle — recognizably yours.

When people say "AI content is generic," they're reviewing the first version. The tool never got a chance to write the second.

Try it yourself · 15 min

Pick a real task from your week. Round A: fresh chat, one-sentence ask, no context. Round B: fresh chat, paste a quick five-item context stack first, then the same ask. Put the outputs side by side: Which could you actually publish? What three decisions did the AI make for you in Round A? And — how long did the brief actually take? That last answer is the one that changes habits.

06 · What to do differently

The brief comes first

1

Run the freelancer test before you type the ask. "What would a day-one freelancer need to know?" Thirty seconds of thinking that changes everything downstream.

2

Front-load the context. The brief goes in at the start of the chat — not drip-fed across six rounds of "no, more casual... no, shorter." Iterating your way toward the brief you never gave is the slowest way to write anything.

3

Judge the tool fairly. If you've been unimpressed with AI output, look at what it was given first. Judge it on briefed prompts, not blank ones — you'll be judging a different tool.

And as you start doing this, notice something: where does all this context actually live right now? For most teams the honest answer is "in three people's heads and a graveyard of scattered docs" — which means every person briefs the AI differently, every time, and quality swings all over the place. Writing it down once, properly, where everyone and every chat can use it — that's Module 5, and it's the payoff this course is building toward.

07 · If you review more than you produce

The reviewer's question

When AI-assisted work comes back generic, the instinct is to blame the writing — or the writer. Usually the problem is upstream. Generic output is a briefing failure before it's a talent failure. Same as it always was with people — we just forget it applies to the machine.

"What context did the AI have?"
Module 2's question · sits next to Module 1's

If the answer is "none, really" — you don't have a quality problem, you have a briefing problem. And if the same context is being re-explained differently by every person on the team, that's your signal the team needs a shared, written source of truth.

The takeaway
The AI isn't generic.
The brief was.
Give it what you'd give a day-one freelancer, and it starts making your choices instead of average ones.
08 · Do the work

Exercises

The before/after ~15 MIN

Covered in the try-it above — run a real task with and without the context stack, side by side. Keep both outputs; they make the case better than any slide.

Where does your context live? ~5 MIN

For each item of the context stack, note where it currently exists: in my head / in a doc (where?) / nowhere.

  • Audience / who it's for
  • Voice / what we sound like
  • Product / offer facts
  • Typical goals
  • What we avoid

Keep this table — it's your personal to-do list for Module 5. If most marks landed in "my head," you've just found out why output quality depends on who's typing.

Reflection

  1. The last AI output that disappointed you: what blanks did you leave, and what did the AI fill them with?
  2. If a day-one freelancer produced that same draft, would you have blamed them — or the briefing?
  3. Who on your team gets the best results from AI? What do they put in that others don't?

Reference card — The Freelancer Test

  • The picture: every fresh chat is a talented freelancer on day one — knows everything about the world, nothing about you.
  • The test: "What would a day-one freelancer need to know to do this well?" Give the AI that.
  • The context stack: who it's for · what we sound like · what we're selling · what should happen · what to avoid.
  • The habit: brief first, then ask. Front-loaded, not drip-fed.
  • The filter: relevant context only — every file eats window space; attach the load-bearing pages, not the whole doc.
  • The reviewer's question: "What context did the AI have?"
  • The line: Every blank you leave, it fills with "average."

Go deeper

Next: the difference between a brief and a wish

Context is what the AI needs to know. But there's a second half to a good briefing: what you actually want it to do. Format, length, angle, what "good" looks like. A fully-briefed AI can still hand you the wrong thing if you never said what you wanted.