I'm ok with AI raising the floor on naming. I'm not ok with it lowering the ceiling.

I've seen what generative AI thinks is good. I know we can all do better.

TW: Cussin’ and bein’ mad. Mostly toward the end…

This is not a criticism of any specific way my peers are using AI.

Many better namers than me have already written deeply insightful things about AI, naming, and creativity—some in favor, some against. I agree with a lot of them, at least a little.

And some namers have built, or are building, what I assume are fantastic businesses around custom AI models and name development.

This is a criticism of how widely available generative AI tools are changing the standards for what a “good” name is—and the process for arriving at one.

I’m a practical person. I know technology is changing, and I work with the tools that best serve the work I’ve (for better or worse) committed my life to.

That said, I’m not a generative AI booster—for the practice or business of naming, or for most aspects of creativity.

While I try (and BOY, do I try) to find ways for AI to improve my workflow and the quality of the work coming out of my studio, I keep running into one specific point of friction:

The gap between assumptions about how fast/cheap/good AI should make my work—and how incredibly mediocre, and sometimes deeply untrustworthy, its outputs are.

I read about AI almost every day.

Most of it, admittedly, is critical: Ed Zitron’s incredible Where’s Your Ed At; Brian Merchant’s Blood in the Machine.

Some of it is neutral to positive: The Deep View.

Some of it is too positive: LinkedIn posts ranging from kind of practical (thanks!) to completely and totally divorced from any reality I recognize.

As of April 2025, there are very few generative AI–based steps in the naming process that I can endorse—at least not without major caveats.

I was (and still am!) working on a series about creative warmups for naming, and I couldn’t shake the voice of my Naming for Everyone students:
How do you use AI for this?
How can I use AI for this?

As I tried to organize my thoughts on what was helpful, I found that for every thing I could share to help you take a step forward, I needed to warn you about the two steps back that often come with it.

So here’s my 2025 take on AI and naming.

---

Want to become a better namer?

The full Foundations of Naming self-guided class series is finally here—take one class or the complete series, and boost your naming confidence.

---

Where the floor and the ceiling meet

On a recent podcast, Jathan Sadowski, author of The Mechanic and the Luddite: A Ruthless Criticism of Technology and Capitalism, framed my challenges with AI so succinctly.

To badly paraphrase him, he talked about how there is a universal raising of the floor when you take into account the human-AI interaction. (In the case of naming, more people can do more of the heavy lift of naming: looking at a lot of names that answer a naming need, more quickly.) The average might get a little higher. But it comes with a drastic lowering of the ceiling. Everyone has the same ability, dictated and reliant upon using ChatGPT, and that strips away the peaks and valleys of human creativity or skill.

The floor raising and ceiling lowering happens simultaneously, as we are asked to accept a lot of “good enough”

Before you ask, I’ve tried a bunch of tools: I’ve tried ChatGPT and Gemini and Claude and Perplexity. I’ve tried out the AI features inside of my design software and project management tools. I use an AI note taker to supplement my handwritten notes sometimes.

Once I even let LinkedIn take a stab at rewriting my draft about a webinar I had coming up and it made me wanna drown myself.

I have tried these tools in a bunch of ways. I will keep trying, because at some point I will get left behind, in some respect, if I don’t. But they do not help me do my work carefully, nor creatively, and I see those two qualities as critical to my own success.

Here is where I see “Good Enough” coming for your work, and where I’d encourage you to watch out

“Good enough” category audits:

Everyone tells me that category audits are a great use of ChatGPT. And sometimes I hear myself agreeing because I don’t want to be a downer.

But here’s the thing.

Either folks are taking ChatGPT at its face value and proceeding with a wildly inaccurate sense of which players make up an industry, or they’re having to painstakingly check each result, to the point where the time saved has been given back to the project in the form of fact checking. Maybe you could have found a direct, authoritative industry source in that time?

Here’s an easy example. A client had wondered about the use of the “-orama” suffix and its current use in market. I thought this would be a great use of ChatGPT to cut down on research time finding examples.

The last name on the list seemed suspicious, so I searched the whole list and found not a single name listed as a major player was real, or existed as framed by ChatGPT.

This is not a productive use of my time.

“Good enough” name generation

I won’t spend too much time here.

I hate the names ChatGPT produces. They’ll probably get better, but right now, they’re barely intern-grade.

If you’re a namer relying on generative AI to do what should arguably the most fun, creative, and rewarding part of your job, there are some questions you can ask yourself. If you don’t know what they are, I’ll ask them for you.

“Caitlin, you don’t understand—we need to look at so many name!”

I do understand. My studio’s projects are largely global. Our lists tend toward 3,000 names per project—and this number rises all the time.

I’ve built tons of custom tools over the years that help me consider a high volume of naming options before deciding what to proceed with. I also get that this step can be helpful at the beginning to “get some initial ideas down.”

Do that if it helps you; if it helps you wade through the initial stage of knowing what direction to take something.

But please don’t stop there.

“Good enough” preliminary screening/desktop searches.

This was the one area where I hoped AI would help most.

I’ve tried dozens of prompts, including asking for citations, to get reliable results when screening for potential conflicts. Every time, one of two things happens:

• It flags a non-existent conflict, OR

• It misses obvious conflicts I plant as a test

Which means my team ends up rechecking every single result—and chasing false positives that were never real. Really fucking annoying.

If you trust in these tools completely, you are likely to be blindsided by someone in the presentation getting suspicious of one name, doing a more traditional desktop search, and asking you why you didn’t catch that, like, Snickers is already a candy bar name.

“Good enough” presentation rationale/making the case for the name

When I hear about namers using AI to develop their rationale for why a name is worth considering, I’m crushed.

THIS IS WHERE YOUR EXPERTISE COMES INTO PLAY.

If you don’t feel like an expert yet, THIS IS YOUR TIME TO HONE YOUR PERSPECTIVE ABOUT WHAT MAKES A NAME A WORTHWHILE CANDIDATE. FUCCKK! SHIT! GODDAMNIT! HELL! CUSS CUSS CUSS!

Don’t outsource this if you really, really want to get good at naming. Don’t let this part of your brain atrophy. Don’t.

You need to be able to talk about why a name should be carefully considered. If your reason starts and ends with something ChatGPT churned out, that argument in favor of the name did not come from within you, you cannot share your process for arriving at that argument live, and you will lose hard-earned trust fast.

“Good enough” name selection

In January, I had a client stop me as I started to explain how I’d like us to start shortlisting a set of names to take into full legal clearance investigations.

He’d taken our naming strategy, pasted it into ChatGPT along with the summary of the names, asked it to rank each name’s fit to the strategy, and then insist we all use that as shortlisting tool.

He meant well! He was excited to use ChatGPT. He loved ChatGPT. It struck him as an efficient use of our time.

Some of you might think, yeah, that IS a super smart use of AI. What’s the problem there?

The problem is that ChatGPT knew he wanted an answer, and it produced one. Now the whole meeting’s trajectory was at the mercy of its ranking.

We were no longer creative human beings, who had embarked on a strategic and thoughtful process, making a decision together about names that had been developed with a ton of careful consideration.

We were forced to agree or disagree with a robot who’d just gotten here, in more ways than one.

AND GUESS WHAT: After some painful reeling-it-back-in, the final name was NOT ChatGPT’s top pick.

EAT MY SHORTS, ROBOT!

Anyway, this was exhausting to write but I feel better and maybe it was exhausting to read and maybe it made you feel…some kind of way.

If you are having experiences with AI on any of the steps above that are precise, reliable, or creative in ways that I have not experienced, course correct me! I’d love to learn. But until then, I still have more faith in books and spreadsheets.

We’ll have some fun together next week. I promise :)

Happy naming!

Caitlin

Want to become a better namer?

The full Foundations of Naming self-guided class series is finally here—take one class or the complete series, and boost your naming confidence.

Download free booklets from the Truth in Branding series on naming and trademarks.

Check out an episode of Big Names in Naming, the podcast where I interview namers about…naming.

All typos are left in to humanize me 🤪