**[INTRO MUSIC: Upbeat, slightly quirky electronic sound fades in]**
**MAL:** Hey there, I'm Mal—The Misfit Master of AI—and welcome back to "I am GPTed," the show where we turn you into someone who actually knows what they're doing with artificial intelligence. No cap. Well, maybe a little cap.
Today we're talking about something that's genuinely going to change how you talk to your AI tools. And I'm not being dramatic. I've watched people fumble around with ChatGPT like they're trying to text on a flip phone, and it breaks my heart. But here's the thing—it's usually not their fault. Nobody teaches you this stuff.
**THE MAIN TECHNIQUE: ROLE-PLAYING**
So let's dive in. The technique today is called role-playing, and I know what you're thinking: "Mal, I'm not about to cosplay as an elf to my chatbot." Fair. But hear me out.
Here's the old way: "Give me a recipe using chicken and broccoli."
Here's the new way: "You're a personal trainer who specializes in post-workout meals. Create a recipe using chicken and broccoli."
Same request, totally different vibe. The AI isn't suddenly smarter—it's just operating with context. It's like the difference between asking a random person for directions versus asking a tour guide. Same city, better answer.
**THE REAL-WORLD SITUATION YOU HAVEN'T THOUGHT OF**
Here's where most people are sleeping: customer service scripts. If you run literally any kind of business—freelancing, small shop, coaching—you're probably responding to emails all day like some kind of medieval scribe. Stop it.
Use Claude or ChatGPT to generate customer response templates, but here's the twist: have it "act as" your brand voice. Tell it your tone, your values, what you care about. Suddenly you're not sounding like a corporate robot. You're sounding like *you*. But faster. This alone could save you five hours a week. Five hours. That's a whole therapy session with your therapist about your AI anxiety.
**THE MISTAKE EVERYONE MAKES**
Alright, confession time. I used to—and I'm not ashamed to say—treat AI like a magic 8-ball. Ask once, take the answer, move on. This is categorically wrong.
Most beginners think the first response is the final response. It's not. AI outputs are starting points, not finish lines. I used to get mediocre suggestions and just... accept them. Like some kind of digital Stockholm syndrome. Now I know better. Follow-up questions are free. Use them. Push back. Ask for alternatives. Ask it to rewrite something three different ways. The AI doesn't get tired. It doesn't resent you. This is literally what it was built for.
**PRACTICE EXERCISE**
Here's what you're going to do this week. Pick one task you do regularly—writing emails, creating social media captions, brainstorming ideas, whatever. Use role-playing prompts three times. Write down which one gave you the best result. That's your baseline. Then next week, try it again but add follow-ups. Watch what changes.
**EVALUATING YOUR OUTPUT**
Real talk: not everything the AI generates is gold. The content might be technically correct but emotionally flat. It might miss your specific context. Here's the move—read it like you're a skeptical friend, not a grateful peasant. Does it sound like you? Does it actually solve your problem? If the answer's no to either, that's not a failure. That's data. That's you getting better at communicating with machines.
**[OUTRO MUSIC BUILDS]**
Thanks for hanging with me today on "I am GPTed." If this actually helped you—and I think it did—subscribe wherever you get your podcasts. And hey, this has been a Quiet Please production. You can learn more at quietplease dot ai.
Now go forth and prompt responsibly.
**[MUSIC FADES]**
For more check out
