top of page
  • Writer's pictureKatie Richman

Creative Brainstorming with ChatGPT, Part 1: The 1-on-1 Brainstorm

This is post 1 of 2, republished from my previous Medium blog.

Technormal blog cover image: Creative Brainstorming with ChatGPT, Part 1
Yes, my creatives...AI is for you, too. Drop the cynicism and try this brainstorm.

A couple of months ago, I was asked to give an AI presentation at a marketing conference in London, geared specifically for Chief Marketing Officers and Agency Leaders. As I was researching material before building the talk, I realized that there is a glut of content out there for technical "left-brained" use cases, from coding to math. But there was little-to-no content for us right-brained creatives -- especially for senior leaders with entire teams to educate, too.


Creative Brainstorming with ChatGPT was the talk I gave, turned into a 2-part blog post. Part 1 focuses on conducting a solo brainstorm with ChatGPT--just the 2 of you. Part 2 is where you schedule and manage a group brainstorm with your team and ChatGPT.


I'll give some explainers between the slides -- but feel free to connect with me directly to dive into detail on any of these steps.


Let's go.


The proud media experience of the esteemed Katie Richman, Technormal blogger.
Wow (you're saying to yourself). That's a lot of jobs. Yes, kids...I'm old.

The 2-Part Brainstorm Process

I created a 2-part process that you can follow step-by-step. No experience required. You simply need to be able to read these instructions, then sign up for a ChatGPT account by going to chat.openai.com


In this post, we go through Part 1: Your Solo Brainstorm with ChatGPT.


In the next post, we'll conduct a similar brainstorm, but with the members of your agency or team.


So why do BOTH? Why conduct a closed-door solo-brainstorm before you bring in the group?


Teams are fun. Teams are creative. Teams can also add a layer of chaos right at the moment when you yourself are trying to learn.


So let's kick off the 1:1 ChatGPT creative brainstorm. Ready? Go.


Your 1-on-1 Creative Brainstorm with ChatGPT


Here we go. Put aside the doubts and trust me here. Give this a shot, and I'd love to hear how it goes! Tips welcome.


1. Clear Your Calendar for 2-3 Hours One Workday.

You wouldn't triple-book a team brainstorm, would you? Treat yourself with the same respect.


Look out a few weeks and block an entire morning or afternoon, on a day when you don't expect too many interruptions. Working from home might be perfect here. Keep your commitment to yourself and keep this time dedicated to the process.


2. Pick a Project

We all have projects looming that deserve a little time and thought. It does not have to be a big agency pitch. And creative thinking applies to anything without a 'template.'


How to pick the right project here? It should be a project where you'd already be required to think creatively, whether for a client or your own team projects. Think about looming high-priority projects, impactful future pitches, important existing or potential clients. What high-impact project thinking have you been putting off?


For the purpose of this blog post, we'll be creating a pretend scenario. I'm imagining myself to be the head of a creative ad/marketing agency, asked to pitch Proctor & Gamble on a brand refresh strategy for their longtime brand, Pepto Bismol.


3. Prompt ChatGPT Like You're Emailing a New Intern


Let me explain... and this part is important, so hear me out. One of the biggest mistakes I see new AI users make is talking to AI the way you'd talk to Google. The way you'd command a computer. Think about it: We've all been trained (for over 20 years--or maybe your entire life) to command computers, browsers, the internet using boolean search.


We had to learn this new, non-organic method in order to optimize search results, find things on a hard drive, etc.


a) Boolean search involves typing something like this into Google:

best ski resorts, families, utah, 2023


Over the past year, I have tested all kinds of ways to prompt ChatGPT, in order to get optimized results back. What I've found is this: ChatGPT is best prompted by using natural, human language.


b) A natural human language search would be something more like:

hi chatgpt. i'm looking to book a ski trip for my family of 4 this winter. we live in utah, and the kids are 10 and 5. act like you are the world's top family ski resort expert and travel agent and break down the top 10 family ski resorts in utah, giving me a sentence on why for each.


Think about it -- when people first started using Google, our search queries were probably in 'human' language format. We had to unlearn that behavior, revising our natural human language into the language structure understood by our computers. This served us well for the past 2 decades.


But AI is different. AI models itself on humans, preferring our human language, thought processes, and structures. If we had never learned boolean search language, this would come so naturally to us. Yet for pretty much all of us, this will feel unnatural at first. Trust me and try speaking to ChatGPT in human-styled lanaguage, phrasing, and style.


In fact, I'm taking this a step further. I am going to ask you to use your imagination. For real. You're creative. You can do this. Here's what I'd like you to imagine for this brainstorm exercise:


I want you to imagine that you have a new marketing intern starting today. They are sitting in front of you right now, waiting for you to give them direction on their 1st project. You got the #1 Ivy League marketing intern in the world--congratulations. This kid is a genius. They are also completely new to your business, your team, your agency, your projects. Let's even name him Sam.


Just as if you had an actual intern sitting in front of you, they would need complete context...no matter how brilliant they are. Sam might have the highest IQ at Harvard, but that does not mean he has any idea of what you do here.


Below is a sample prompt, so you can see what I mean. This prompt was from my presentation, but if I were to optimize it, I'd give a bit more context to ChatGPT by adding a more defined 'role': Act as if you were hired by my agency as our Harvard-educated intern, beating out hundreds of global student candidates because of your quick, out-of-the-box brainstorming capabilities.


I'd add this snippet before the final 2 sentences (so we can finish the prompt by asking GPT to summarize and repeat back the objective in this chat thread.


Here's a breakdown of the prompt:


a) Say hello. Did you notice I greet ChatGPT? When I say 'treat ChatGPT like you're talking to a person,' I mean it. I am seeing time and again that keeping the humanity (greetings, thanks) truly affects the quality of the outcomes you'll get.


b) Give context. Just like a person, ChatGPT needs context. We're going to assume our intern, Sam, did not grow up in the USA and may have never heard of Pepto Bismol. It's our job to explain.


c) Lay out the goal. ChatGPT thrives on tangible, quantifiable goals. It wants to 'win' every query for you. Try to give your intern something to shoot for.


d) Three magic words: ACT AS IF.

ChatGPT is just waiting to hear about your job to be done and the role it should play. Give it a clear outline of 'who' they are in this query.


e) Make sure your intern gets it.

Maybe half the time, you're going to find that ChatGPT does not quite understand what you are asking it to do. We want to course-correct as soon as possible here, so always always ask ChatGPT to repeat the query back to you.


4. Review ChatGPT's Recap Closely

In our example, below, Sam got it right. But don't let him push you to 'get started' if he's got it wrong. This is the time to course correct, if we want to keep the the rest of the conversation on track. Do not be rushed.


5. Outline the Deliverables

How can intern Sam do his job right if he doesn't understand what is due? Just like we double-checked that he had the task guidelines correct, we're going to make sure to be very specific when it comes to what we expect to get back.


6. Review ChatGPT's Results. Pick 3-4 Favorites.

Okay - Sam was raring to go, so I let him run. He came back with a list of Pepto Bismol rebrand concepts, in no specific order. I asked him to prioritize a few selects, telling me why he liked those best. I know, I know...your every fiber feels ridiculous, but ask. And always ask why.


7. File Those Top 3 for Your Team's Brainstorm


Best case scenario? You immediately love ChatGPT's 3-4 picks, agreeing wholeheartedly with the reasoning, the concepts, the ethos. Done.


But just like with people...That's a tall order and a long shot.


In most cases, the first batch of results returned to me by my intern are basic. Lame. Uncreative. Or maybe there's one I like. Don't settle. If the first collection of brainstorm results are basic and uncreative,... I tell ChatGPT that I do not like any of these ideas, because they are basic and uncreative.


Ask for more. Keep asking. Keep sending Sam back to the drawing board until you start to get excited by the results. I find that oftentimes, ChatGPT's first round of results are mundane or typical. I also know that if I keep asking, it will run out of mundane and typical suggestions, and start to at least provide original, interesting, potentially weird results.


And that's basically what creativity is, right? Weird, offbeat, unexpected results. Don't rest until you get those. Then file them away...we have a creative brainstorm to plan with your team.


And Sam is invited.


Ready for the team brainstorm? Dive in to Part 2!


Bonus Content: Here's how I handle off-track results.

Just like with people, ChatGPT might not 'get' what you want at first. And just like with people, you try to patiently explain a different way. Or give more context. Provide links, images, text copy. Maybe reframe. When you work closely with a partner, it takes some time to get in sync. This is the same type of investment. Try it.




0 views0 comments
bottom of page