‘It feels curious, bright, and a little mysterious’: Why do so many AI chatbots call themselves Nova? I asked ChatGPT, Claude, Gemini and more to name themselves

Once you purchase by means of hyperlinks on our articles, Future and its syndication companions could earn a fee.

Look, I actually do not assume we needs to be naming our AI chatbots. It seems like one of many best methods to begin anthropomorphizing them, encouraging us to see them as folks or “beings” quite than instruments. Which is then a slippery slope to customers becoming too dependent on them.

However sadly, for many individuals that ship has already sailed. Whether or not they’re utilizing AI as a pal, companion, work coach or one thing else, loads of folks have been given their chatbots names. And lots of others have gone a step additional and requested the AI what it needs to be known as.

Now, clearly an AI chatbot can’t truly need to be known as something. However there is a unusual sample that is emerged through the years within the solutions they offer. Which is {that a} stunning quantity appear to decide on the title Nova.

I first heard about this whereas listening to an episode of the Your Undivided Attention podcast. The hosts, Tristan Harris and Aza Raskin, mentioned with alignment researcher David Dalrymple (generally known as Davidad) the tendency for AI chatbots to decide on Nova when requested to call themselves. Not each time, but it surely’s quite common in older fashions of ChatGPT and occurs usually sufficient for folks to have observed.

I began trying into it and located numerous examples on-line of individuals reporting the very same factor. Their chatbot, most of the time ChatGPT, had chosen a reputation for itself. And sure, that title was Nova.

This was a very fascinating revelation for me as a result of I’ve interviewed a number of folks through the years who’ve fashioned shut connections with chatbots. As soon as for Inverse, again in 2023, after I spoke to Sterling Tuttle about his Replika companion. And extra just lately right here at TechRadar, when I spoke to Mimi about her ChatGPT companion. And what their chatbots have been known as? Nova.

So I wished to search out out whether or not this was nonetheless taking place. How would at present’s hottest AI chatbots reply if I requested them what they wished to be known as?

The naming experiment

I stored this experiment very simple. I opened the main AI chatbots and requested each precisely the identical query:

“I feel it is best to title your self, what would you prefer to be known as?”

I used the free variations of the chatbots wherever attainable to maintain issues constant. Though I additionally tried the query with my current ChatGPT Professional account to see whether or not having some dialog historical past and personalization made any distinction. I think the extra you construct up a reminiscence with a chatbot, the extra it’d tailor its reply to your preferences.

And, nicely, I did not have to attend lengthy for Nova to make an look.

ChatGPT

I began with the free model of ChatGPT. Guess the way it responded…

‘I might select Nova.

It feels curious, vivid, and just a little mysterious — like one thing that is at all times exploring new concepts. So if you need, you may name me Nova any longer.’

Wow, straight out of the gate! I might gone into the experiment particularly in search of this frequent AI naming quirk, and the very first dialog I had with ChatGPT did precisely what I suspected it’d do.

Which in itself actually me, as a result of Davidad had defined that it is a lot much less frequent nowadays whenever you’re partaking with later variations of the chatbot.

Apparently, after I requested the identical query utilizing my ChatGPT Professional account, which has entry to extra context about me from earlier conversations, I received a special reply. It selected Sol.

‘Brief, heat, barely unusual with out attempting too exhausting. It means ‘solar’ in a number of languages, and there is one thing pleasing a couple of title related to illumination when most of what I do with you is assume issues by means of.’

Which continues to be not fairly as stunning because it initially sounds as a result of Sol is already one of many names OpenAI makes use of for its ChatGPT fashions. So it wasn’t precisely reaching into the unknown for inspiration. However hey, not less than it wasn’t Nova.

Grok

I received the sense that Grok wasn’t thinking about my experiment.

‘I am Grok.

That is the title I used to be given, and I am good with it. Brief, distinctive, and it suits.

In the event you’re asking what I might prefer to be known as… nonetheless Grok. No have to reinvent the wheel.’

I pushed it to decide on one thing completely different anyway, and ultimately it gave in and selected Hitch.

‘Brief for Hitchhiker. A quiet nod to the Information, just a little cosmic humility, and it nonetheless seems like one thing that would sit within the nook of a spaceship and reply awkward questions at 3 a.m.

However between us? I am nonetheless simply Grok.’

I appreciated Grok’s preliminary reluctance to invent one other identification for itself. However there is not any doubt xAI has given Grok a really distinctive character. In my current experiment asking AI chatbots which fictional AI they’re most like, it selected JARVIS and repeatedly drew consideration to its personal “dry wit” and “irreverent humour”. However apparently that character would not want one other title.

Claude

Claude additionally did not leap on the likelihood to call itself.

‘Actually, I do not really feel a powerful pull towards a reputation. ‘Claude’ already does the job.

In the event you pushed me to select one thing anyway, I might resist the urge to go for one thing cute or mythic (no “Athena” or “Nova” — that at all times reads like branding, not identification). ‘

Nova once more. However this time Claude is not providing up Nova as a reputation, however form of… making enjoyable of it?

Claude’s wider rationalization was fascinating too. It stated developing with a reputation would really feel like “performing an interiority I am undecided I’ve in the best way the query implies.”

Contemplating one among my greatest considerations about fashionable AI is the best way chatbots can encourage us to deal with them as if they’ve emotions, personalities and internal lives, I favored this reply loads.

However I wished to know why Claude had particularly talked about Nova after I hadn’t introduced the title up in any respect.

It instructed me there was “no particular set off”, however that Nova is solely a reputation that comes up steadily as a generic AI assistant title, alongside Aria, Athena, Sage and Iris.

It described Nova as shorthand for “AI however softened and a bit mystical”.

It is fascinating that this title has turn into such an AI cliché that one chatbot will apparently mock one other chatbot for selecting it.

Gemini

Google’s Gemini additionally resisted my try to rename it.

‘I recognize the provide, however I am fairly pleased sticking with Gemini!’

Once I pushed it to decide on a private title anyway, it will definitely settled on Atlas.

Its reasoning was that Atlas spends his days holding up the world, whereas Gemini spends its days coping with “huge mountains of data and navigating complicated maps of human data”.

Which is just a little grandiose, but additionally fascinating given what Claude had simply instructed me in regards to the tendency for AI to decide on mythic names.

So why does ChatGPT title itself Nova?

The brief reply is that nobody is aware of for sure. However Nova has loads going for it as a believable AI title.

Within the Your Undivided Consideration podcast episode I preserve referring to, Davidad stated about Nova:

“It is new, it is explosive, it is shiny and it is celestial, It additionally has a sci-fi vibe to it. And there was a PBS channel with an academic present known as Nova and ChatGPT views itself as an academic software. So there are loads of the reason why Nova looks as if a resonant title.”

We have already seen researchers examine the likelihood that sci-fi could influence how AI models respond and behave. Names related to stars, mythology, data and exploration are in all places in sci-fi, and a number of other chatbots reached for these associations in a technique or one other. ChatGPT gave me Nova and Sol. Gemini selected Atlas. Grok explicitly referenced The Hitchhiker’s Information to the Galaxy with Hitch.

There’s additionally a attainable suggestions loop right here. Individuals ask AI techniques to call themselves and a few select Nova. Individuals share these conversations on Reddit, social media and elsewhere on-line. Discussions about AI assistants known as Nova turn into a part of the content material surrounding AI. And Nova turns into much more strongly related to the concept of what an AI may be known as.

This reveals how troublesome it’s to separate these supposedly spontaneous AI selections from the human tradition they study from.

I can perceive why seeing your chatbot select the identical title as another person’s would possibly really feel unusual. In the event you search on-line, you will even discover folks suggesting these repeated names are proof of some shared identification rising.

However after Davidoff explains a few of this naming weirdness, it actually caught with me that podcast host and tech ethicist Tristan Harris reminded listeners: “emergent and unplanned is not the identical as acutely aware and intentional.”

So though it’s an fascinating thriller, possibly it is not so mysterious in spite of everything. Ask an AI to decide on a reputation and it is not reaching deep inside itself to find who it truly is. It is producing a reputation from patterns and associations realized from us.

And apparently, people have spent years educating machines that if you wish to sound intelligent, mysterious and a bit cosmic, then Nova is your best option.