To anthropomorphize, or not to anthropomorphize? That is the question for marketing departments tasked with branding generative artificial intelligence.

There is of course the very nonhuman-sounding ChatGPT. And then there is the very human-sounding Claude, from an AI startup appropriately called Anthropic.

Google has gone the Shakespearean route, naming its AI chatbot Bard. So, what’s in an AI name? Well, considering the billions of dollars tech companies have invested in these products, a lot.

“You don’t realize how much consternation goes into this naming thing,” said David Ferrucci, head of the AI company Elemental Cognition.

Ferrucci previously worked at IBM and led the development of Watson, the supercomputer that destroyed Ken Jennings and the rest of the meat puppet competition on “Jeopardy” back in 2011.

He remembers rounds of meeting with the marketing team, debating several different names. Watson was an early leader, mostly because it shared a name with an early 20th century IBM president.

But there was another option.

“People were talking about Sherlock,” said Ferrucci.

Sherlock, of course, is the titular star, the brilliant one, the guy who actually solves the murder.

And that’s not really the role Ferrucci envisioned for AI.

“I liked the notion that computers were helping,” said Ferrucci. “Helping humans make better decisions, and helping humans collect the facts and things like that.”

Watson was impressive, but this new wave of generative AI tools can do even more, like writing legal memos or winning art competitions. Which presents some dilemmas for branding teams: lean into the power of the tech, or reassure anxious customers that you’re just here to help them, not replace them.

“So on the one hand … Microsoft has Copilot,” said Katy Steinmetz, project director at Catchword, a branding and naming agency. “It’s Copilot, not Autopilot.”

And on the other hand, you have consulting firm McKinsey’s AI service, QuantumBlack, a name that says, “We’re like Skynet from the ‘Terminator’ movies, only more terrifying.”

Steinmetz said the decision to use a human name can make your AI sound more relatable and friendly, but it also raises expectations for performance.

“It invites you to think about a chatbot as an entity, as an individual, as someone who can understand nuances a human could. And that could be overpromising depending on the company,” said Steinmetz.

She said companies that are christening their AI with human names are increasingly going gender-neutral or male. Apple’s Siri and Amazon’s Alexa were criticized for reinforcing sexist tropes of women as helpers and assistants.

The male names have to strike the right tone, though. You’re probably not going to see GenghisGPT anytime soon.

“I think you look at a name like Claude, from Anthropic. I would say it’s a kind of gentle male name,” said Steinmetz. “If we know any Claude, it’s probably Claude Monet. Maybe Claude Debussy. Nice music, nice painting, maybe a little too French for you, but it doesn’t at all sound threatening.”

There’s another dimension to choosing a human name tech companies have sometimes neglected. The actual humans.

“It just got a lot easier at some point to go by Lex,” said Lex Vaughn, editor in chief of The Needling, a Seattle humor website. “There’s a certain point where you stop wanting to hear the same joke about your name.”

Vaughn used to go by Alexa. Until stuff like this started happening:

“I’ve had friends Facetime their kids while visiting me, and they think I am the Alexa, from the device,” said Vaughn.

Alexa switched to Lex a few years ago, and it’s working for her. Well, at least for now.

Lex is also the name of a new AI writing assistant. Human Lex won’t be using it for her job anytime soon.