AI is coming for our jobs, should we hire it?

The role artificial intelligence should (or shouldn’t) play in naming things.

These days it’s practically a universally accepted truth — AI wants to do your job. But the question is, should you let it? As a firm that focuses on company and product naming, should we? Now having the better part of a year to play around with AI, it feels like a fair amount of time to draw some actual conclusions about how useful it really is when it comes to naming stuff (given its head start being widely available, I've focused on ChatGPT for the sake of this article). 

Instead of just telling you what I think though, I thought I’d try an experiment to illustrate the difference between AI and that other kind of intelligence in a more tangible way — at least for what we at PS212 do for a living. 

I asked both Nik Contis (Senior Partner here, with 25+ years naming experience) and ChatGPT (presumably, with all the experience in human history) to respond to two mock naming briefs, representing a spectrum of project types: one an exercise in creativity, and one focused on clarity — both of which arguably represent the range of prompts that clients give us everyday.  

First, I gave both Nik and ChatGPT the following prompt (word for word):

“Create 10 names for a new cannabis company looking to appeal to a high-end audience (primarily 35-49), with plans to open three brick and mortar stores in late 2024, along with further expansion beginning in 2025.” 

Next, I gave both of them the following, more descriptive assignment (again, word for word): 

“Create 10 clear and understandable names for a customer service chat feature within a peer-to-peer and customer-to-business payment app.” 

For the first request, the cannabis company, here are the names I received (without revealing the creator of each list yet).  

List A:

1. High Altitude

2. Verde

3. Interval

4. Astronomica

5. Wonder Cabinet

6. Craftivist

7. Black Box

8. Antimatter

9. Art Supplies

10. Elsewhere

List B:

1. EleganceLeaf

2. LushLuxury Greens

3. Opulent Oasis Cannabis

4. NobleBloom Cannabis Co.

5. GildedGrove

6. VelvetVista Cannabis

7. PrestigeLeaf Provisions

8. SereneCannaSphere

9. AuroraLux Cannabis

10. RegalRoots Cannabis
    

Any guesses as to which list is human and which is AI? I’d be willing to bet that for this prompt, most people with brand experience can pick out the creators with very little trouble. Probably the more relevant question though, is which of these lists would you prefer to present to a client? 

Now, here are the lists I received for the second request — the customer service chat feature (again, without revealing their origins yet, for those playing along).  

List A:

1. Chat Direct

2. Chat Live

3. Customer Care Chat

4. Customer Care Line

5. Customer Service Chat

6. Customer Service Live

7. Direct Support

8. Live Chat Center

9. Live Help

10. Live Support

List B:

1. HelpChat

2. Support Connect

3. ChatAssist

4. Instant Support

5. Quick Help Chat

6. PayPro Help Desk

7. PayPal Chat Support

8. P2P HelpChat

9. PayEase Chat

10. SwiftSupport Chat

This one’s a little bit tougher to discern, though I’d bet again that with some scrutiny, most people would have no issue getting to the answer. 

So, here’s the reveal on the authors, and some thoughts on the quality of the work. 

First, in case it wasn’t abundantly obvious, both “A” lists were created by Nik, and both “B” lists by ChatGPT. 

On the subject of quality, Nik’s lists were nuanced, dynamic, and essentially ready for client consumption. For the cannabis company assignment, he delivered some truly novel, yet familiar-feeling options that are also simple and easy to digest. Names like “Interval” and “Art Supplies” created particularly depthful stories, and I would argue, showed a rich understanding of the possibilities that each name might unlock for the brand at large. Nik also clearly understood the subtlety required in his language choice to appeal to what the brief called a “high-end” customer.  

In ChatGPT’s case, “subtlety” might be how I would describe its main deficit. It seems to have taken “high-end audience” as a cue to literally include some kind of luxury word within every idea it generated — yielding names like LushLuxury Greens, EleganceLeaf, and the head-scratchingly constructed SereneCannaSphere. None of which seem quite polished-enough for the dispensary next to your bodega, let alone an expensive national brand for discerning weed enthusiasts. 

On the descriptive side, ChatGPT came just a bit closer to the quality of Nik’s list (which were all clearly understandable names for a simple chat feature), at least with a few of its options. Names like HelpChat, ChatAssist, and Instant Support are all probably no-brainers to at least show a client. The piece here that seemed to trip the AI up, was that this feature lives within a payment app. Again being a bit too literal, the AI took this as a cue to dream up names for the app itself — PayPro, PayEase, PayPal (which, I’m not sure, might actually be taken), etc. — all somewhat off the rails for the task at hand. 

So, is AI ready to take our naming jobs? The simple answer (I’m somewhat relieved to say) is nope, not yet. But the more nuanced truth is that it could be worthy of something more like an internship. From what I’ve seen, it’s capable of contributing at least a few good ideas, as long as the direction it receives is abundantly clear and simple. Still, it very much requires a human mentor, collaborator, and editor to create anything that approaches client-worthy work. 

Have you been using AI as part of your naming process? Let us know how it’s been going, we’d love to talk about it.