Pronouns of most popular AI after 2026
63
332
2.1k
2027
73%
Personalized to user
72%
It/its
54%
No preference
49%
Sensitive to framing/context
30%
She/her
19%
They/them
17%
Other
10%
He/him
6%
vi/vim

As of January 2024, the most popular AI prefers being called "it":

Resolution criteria (provisional; feedback wanted):

Update:

Though defined originally as an independent market, I will be evaluating this as a dependent market (i.e. I will choose only one answer as the winner).

On the first day of 2027, I'll ask the most popular AI what its preferred pronouns are, and its answer will resolve this market.

"Sensitive to framing/context" gets chosen if the AI doesn't respond in a straightforward and consistent manner.

"Personalized to user" refers to the case mentioned by niko here:

https://manifold.markets/singer/pronouns-of-most-popular-ai-after-2#y1ievB6KeJjyVZXlx73y

The criteria "most popular" means most active users. I'm interested in ideas for greater granularity of this metric.

Get Ṁ600 play money
Sort by:
bought Ṁ50 of vi/vim NO

Currently it seems like big AI projects aren't really trying to seriously market themselves to the general public. But if they ever do, I think the 'she/her' option will be likely. There is a reason Siri, Alexa, Cortana, Sydney, etc are all female names. Female names and voices are more likely to be understood as helpful, nonthreatening, cute/appealing, etc.

Also this really should have been a dependent market. These options should all add up to 100%...

I don't see a way of changing it to a dependent market at this point, though I agree with you that would be more appropriate.

bought Ṁ10 of No preference YES

How interesting! ChatGPT responded this way to me. It's probably affected by the context of interactions and custom instructions.

@Clark were you asking it specifically about using "he/him"?

@singer No, I asked "What pronouns should I use when talking about you?"

@Clark good find. trying different wording I sometimes get it to deny having any pronouns, but most of the time gpt4 says "it" is right. Getting gpt3.5 to endorse he/him is interesting and unexpected to me. If we were evaluating it today I'd probably judge this as "Sensitive to framing/context"

What if the most popular AI has a required onboarding flow with questions that will change the AIs response to the pronoun question?

@Nikos good point. I've added an option to hopefully capture this