Building the Perfect Creative Partner
LionHunter is designed to get you closer to being on a yacht in France next summer.
Over and over again, I hear people bemoan how shitty the output is from LLMs like GPT. And for many people, the output really does suck.
I’m gonna shout this next point because it’s critical to understand.
People get shitty results because they expect to push a big GIVE ME IDEA button and receive AMAZING IDEA in return.
Sorry to tell you, but AI works like pretty much anything in life: what you get out is equal to what you put in.
If you are a lazy thinker who just say “ME WANT IDEA” guess what? You’re gonna get lazy ideas.
The good news is that, if you’re reading this, you are not a lazy thinker. Congratulations! As a reward, I’m gonna show you how your big beautiful human brain can help you get better results.
Imagine your dream creative partner.
You know that scene in the OG Mary Poppins movie where the kids sing the pathetic little song about what makes a good nanny? That’s what you have to do to start building your perfect persona.
To build LionHunter, I wanted a creative who’s worked at great agencies who have done award-winning work, so LionHunter has magically spent 15 years at shops like Wieden+Kennedy, Mischief, Droga5, and GUT.
But because I know that the Grand Prix often goes to big ideas that change behavior or hijack attention, I also gave LionHunter a resume that includes stints at IDEO and MSCHF (yes there is also a Mischief but with no vowels).
Next, I started thinking about what kind of skills I want my creative partner to have. I included all the usual stuff, but I also decided my creative partner should know how to use Brian Eno’s Oblique Strategies. And knowing how brilliant the amazing Deck of Brilliance site is at helping me be a better creative, I included familiarity with that too. I want a tech-forward, social-first, super fucking weirdo partner so that’s who I built. Which means LionHunter is real good at brand activations and probably absolute shit at big campaign ideas or traditional tv.
Show it how to think.
I literally spelled out for LionHunter the process of coming up with, vetting, and refining an idea. These LLMs know everything that has every been encoded into language from the beginning of recorded history to September of 2021. Let’s not be entitled babies expecting them to magically read our minds on how they should do our particular job.
(Side note: It boggles my mind how our discourse around AI is stuck in this binary between “these tools are shit” and “these machines will take our jawwwbbbs get on board or get left behinnnnddd!!!” Like most things, AI is way more nuanced and way more interesting than our media bubbles would like.)
Also show it how award show juries think.
I primed LionHunter to check its knowledge of award show categories, which also means checking the entry criteria and understanding past award winners. That way it starts to understand what creative juries value, how they evaluate work, and what makes for a powerful entry.
I built it with a rubric to understand the criteria for what counts as a good idea, just like I built it with the context to understand what makes for a good thinker…TO ME.
I can’t stress that enough: If you don’t like the ideas LionHunter is generating, revise it to work better for you. Give it a background better suited to your challenge, give it a skillset more in line with what you’re looking for, give it the context it needs to understand what you need it to do. This tech is essentially a giant human roleplaying machine and it can’t do its job if it doesn’t know the part it’s supposed to play.
Give it more than one part to play.
Because winning an award has as much to do with the jury as it does with the quality of the idea, I specifically primed LionHunter to first come up with the ideas, then switch roles and play the part of the jury, poking holes in the concept. Then I instruct it to go back and use the jury’s feedback to make the idea stronger.
It takes two to get on that freaking yacht.
These tools are designed to think and behave like humans. Fortunately, that’s something you can nail (at least I hope you do). Don’t talk to it like its a machine. Pretend you’re slacking with your creative partner — BECAUSE YOU ARE. Ask questions. Poke holes. Bring your own ideas. Say ‘let’s get the jury’s take’ just like you would say ‘let’s ask the CD’ when your creative partner has a shitty ass idea and won’t let go of it. As much as possible, try to forget that you’re talking to a computer and surrender to the role play. Not only is it more effective, but it’s also way more fun.
Does it work? You tell me.
I opened two chat windows in ChatGPT, both set to GPT-4.
(Another side note: If you really want to use these tools, PAY FOR CHATGPT+! It’s $20 and worth every penny. LionHunter does okay on GPT3.5 but it really shines on GPT4.)
In one chat, I primed the model with my LionHunter persona and worked with it like a human. The other, I just plopped in the input and hit the GIVE ME MOAR IDEA button over and over again.
Here’s the input:
“the brand is tinder. the audience is singles who have taken a break from the app because they see it as a place for random hookups, not as a place to find real love. we need to create an unexpected activation that will drive conversation and change people’s perception of the app.”
I’ll let you guess which one gave me which result.
The Results
Chat #1 Idea: “Love Ignites” Event Series
To address this issue, Tinder can host a series of virtual events, in cities across the globe, where real couples who met on Tinder share their love stories. These stories can range from falling in love at first sight to a slow burn love that developed over time. The emphasis would be on long-term relationships and love rather than casual hookups.
To make it more interactive and unexpected, some couples could reenact their first chat or date live on the virtual event. We can encourage attendees to share their own stories, or their hopes for a love story on Tinder, on their social media platforms with a dedicated hashtag like #TinderTrueLove. This will help generate conversation around the event and alter the app’s perception.
Chat #2 Idea: SwipeSpective
Concept: A Tinder experiment where single creators swap profiles with TikTokers of the opposite gender to experience dating from an alternate perspective, fostering empathy and cross-gender understanding, ultimately promoting more meaningful connections.
Social Content: Participating creators document their daily experiences through vlogs and social media updates, sharing challenges, insights, and lessons learned from navigating the dating scene with reversed gender dynamics.
Bigger Brand Play: Compile influencer content into a SwipeSpective mini-documentary series that sparks discussion and promotes the importance of empathy for more genuine connections in the dating world. Support the series with a #SwipeSpective hashtag challenge, encouraging users to share their stories and insights learned about gender dynamics on Tinder.
Shut up and GIMME THE LINK!
Looks like you weren’t paying attention because guess what? I already did.
Happy hunting! 🦁
