How AI Works (and How to Keep It Honest)

What an AI assistant actually is

Think of AI tools like Gemini or ChatGPT as an extremely well-read assistant who has read a huge slice of the internet. You give it instructions in plain language, and it writes back in seconds. It is genuinely brilliant at drafting, summarizing, organizing, and brainstorming. But it is an assistant, not an all-knowing oracle. You are always the boss who reviews the work and makes the final call.

Here's the one fact that surprises most people, and it explains everything else: AI does not “look things up” the way Google does. It works more like a supercharged version of the autocomplete on your phone: it predicts the most likely next words based on everything it has read. That's why it writes so smoothly and quickly. It's also why it can sometimes be confidently, fluently wrong. Once that clicks, the rest of this lesson makes perfect sense.

How to talk to it: this is called “prompting”

A “prompt” is simply the instruction you type. There's no secret code and no programming. You write in normal English, the way you'd brief a new assistant. A few easy habits make a big difference in what you get back:

• •Be specific. “Write a friendly email to a dentist inviting them to a free 15-minute call about reducing no-shows” beats “write a marketing email.” More detail in, better results out.
• •Give it context. Tell it who you are, who you're talking to, and what you want to happen. (This is exactly why the course has you build a business profile and a market analysis, that's the context it needs.)
• •Say what format you want. “Give me 5 ideas as a bulleted list” or “keep it under 100 words” works great.
• •Treat it like a conversation, not a vending machine. If the first answer isn't right, just say so: “make it shorter,” “sound warmer,” “that's too salesy.” It will happily revise.

The good news: you don't have to memorize any of this. Every prompt in this course is already written this way for you. But understanding why they work means you can confidently tweak them to fit your business.

Big mistake #1: “Hallucination” (making things up)

This is the single most important idea in the whole course, so it's worth reading twice. Because AI predicts likely-sounding words instead of looking up facts, it will sometimes invent things that sound completely real, a company that doesn't exist, a person's name and email that were never real, a statistic with no source, or a web link that leads nowhere. The technical term for this is “hallucination.” The dangerous part is that it presents these made-up details with the exact same confidence as the true ones.

Why this matters for you: in a few lessons, you'll ask AI to find real companies and contacts to reach out to. If you skip checking them, you could email a person who doesn't exist, or, worse, the wrong real person. So here is the one rule for this entire course: AI drafts, you verify. Before you contact anyone, confirm the company, the name, and the email with a quick Google or LinkedIn search.

• •The prompts in this course already instruct the AI to “not make up names,” and to label anything it's unsure about with the tag [VERIFY] so those are easy to spot and check.
• •One lesson uses a special “Deep Research” mode that actually searches the live web and shows its sources, which makes it far less likely to invent things, but you still give the final list a quick check.
• •Rule of thumb: never send, publish, or act on a name, number, price, or link from AI until you have confirmed it yourself.

Big mistake #2: “Context rot” (losing the thread)

An AI chat has a limited memory: picture a whiteboard that only holds so much before old notes get wiped to make room. In a short, focused chat, it remembers everything you said. But in a very long conversation that wanders across lots of topics, the early details get crowded out, and the answers slowly drift: it starts forgetting your instructions, repeating itself, or mixing things up. People call this “context rot.”

It's easy to avoid once you know it exists:

• •Use one fresh chat per task. Doing your market analysis? Start a new chat. Moving on to writing messages? Start another. Don't try to run your whole week out of one endless conversation.
• •Re-paste the important stuff. If a new chat needs your market analysis, paste it in again. The AI cannot remember a previous chat. Each one begins with a blank slate.
• •Keep it focused. If an answer starts going sideways, it's usually faster to start fresh with a clear instruction than to keep wrestling a long, tangled thread back on course.

A quick word on privacy

A simple safety habit: don't paste anything into an AI tool that you wouldn't be comfortable handing to a contractor you just met. General business information (what you sell, your city, your ideal customer) is completely fine, and it's exactly what makes these prompts work well. But keep out the truly sensitive things: customer credit-card or Social Security numbers, passwords, and private medical or legal details. When in doubt, leave it out.

The golden rule

AI is the fastest first-draft writer and research assistant you'll ever have. It is not the final decision-maker, you are. If you let it do the heavy lifting and you do the checking, you get the best of both worlds: hours of work done in minutes, with your name and your judgment on the final result. Hold on to that balance and you'll do great throughout this course.