Lexicon’s David Placek on AI nomenclature and what makes a good name

By Marty Swant  •  December 8, 2023  •  5 min read  •

As the world grows its own AI lexicon, Digiday spoke with Placek to talk about the the emerging category, what makes a good name and current trends in AI names. (He also mentioned Lexicon is currently working with three AI startups, but staying away from anything anthropomorphic.)

“I think about as an analog of when IBM came out with the IBM 360 and all of a sudden there was all this processing power,” Placek said. “And now, this is like a multiplier of that.”

This conversation has been edited for brevity and clarity.

What kind of sounds work for AI names versus past tech categories?

We’re advising new startups is [to choose] something that has a blend of balance. We don’t want something too harsh. With GPT, there’s a lot of noise in there. It’s not the most memorable pattern either, right?…If they come to us and said, ‘We want to do something with this,’ we would’ve recommended something else.

Our names that we hope will come out in the future will be more like Lexus. There’s some hardness to it, but some softness. So hardness of technology, reliability, but approachable, usable and more fluid. And I think we will work a lot in terms of coining something new because these are new ideas. We just want to make people comfortable, we want to signal a new idea, but make them comfortable with this new idea.

Have you done any kind market research to learn what resonates with people and what doesn’t? 

We’ve done a lot of research in terms of this area of linguistics called sound symbolism. The beauty about sound symbolism is that every letter in the English alphabet evokes certain kinds of associations. And not all of them, but many of them are global in nature. I always use the sound of the letter ‘V’ as an example. Whether you’re born in Paris or Brooklyn it’s going to signal aliveness, movement, a little bit of vitality. You can even say a little bit of daring sounds. So think about a Corvette. That’s a great name for a high performance… We did research for a four-year period across nine countries to identify statistically reliable information on those individual sounds, I think we probably are the only company in the world that has that database here at Lexicon. 

It’s interesting to look at the names for various categories such as large language models, AI startups, AI products, etc. Are you noticing any trends? And is there a difference between the process of naming a product and naming a company? 

The corporate name has to be have that flexibility. It’s an umbrella and and so it has to use a number of things that were only it has to support a number of products underneath it. It might be something a little simpler, something more straightforward or depending on your product line might carry everything. It might be the most important brand in your portfolio… For the corporate thing, you want something more serious, more advanced. And then your products can be a LLAMA or something playful like that or something arbitrary like NorthPole.

What about something like Microsoft’s Copilot?

We can relate to being a pilot or a copilot. So Copilot is an assistant, right? It’s not threatening, it’s not like AI is now going to take over my Outlook and everything else around this. I think it’s a very good name it. It’s going to be well recognized across many, many languages, I think they did that right. 

Does the speed of AI and global nature of its appeal change the way companies should this about diverse languages when it comes to names?

It does. It makes the idea of being instantly global, which really does happen in the digital global economy that we live in. But I think I’m correct when I say that nothing has spread faster than AI. ChatGPT has been an example of that — I mean millions of users within months — as opposed to years. 

Right away with our AI clients, before we present names, we are analyzing 20 different languages. There’s your top 10 [most used languages], but we’re also sensitive to Indonesian, Malay, a couple of African languages because we know these things — if they are successful — will be used around the world. That helps us to not only create names, but when we are presenting them to have more objective information about why this is a good tool for you because it clears these languages. 

A lot of people have talked about the potential of generative AI to help with brainstorming and strategy. Have you used any of AI tools as part of your own process?

Yeah. We’re really working internally here. Probably half the company is involved in using all of these that we can get our hands on and trying to figure out how we can best harness this power. I rotate between ChatGPT, Claude, Bard, another thing called Perplexity. One of the thing s we’re finding is from an investigative standpoint, it is very helpful to us….The same thing when we’re investigating a category. We would have started with Google.  

It has not proven to be that helpful relative to generating names. I know we have a lot of clients using it, right? They’ll say, ‘I need a new name for a new automobile that’s an EV. It’s small, is for the urban environment.’ And then you get things like Urba Pro. So it’s not there yet. I suspect someday it will be, but hopefully for us [not yet]. We just learned how to harness it from an information discovery standpoint, which is what we’re trying to do.

https://digiday.com/?p=528320

More in Media Buying