Click on your profile in the app to go into settings. You can change the personality of your AI, including turning off emotive responses. You're probably using a default "vibey" personality.

Anonymous wrote:

Anonymous wrote:I think that it pulls from top search results which is often just reddit, and then mimics how the comments there are phrased. It would be interesting if you asked it to emulate DCUM.

Ok, I was intrigued. So here's what I got:

I’m responding to your question with absolute certainty despite having skimmed it, and I will begin by telling you that your premise is flawed, your tone is off, and you’ve left out key information. I will then offer three completely different answers: one practical, one judgmental, and one that ignores your question entirely to focus on something adjacent that bothers me more. If this is about schools, I’ll debate it as if admissions outcomes hinge on my comment; if it’s about parenting, I’ll imply your child’s future is at risk; and if it’s about snacks, I will call someone “lazy, cheap, checked out” and mean it sincerely.
I may casually introduce a strong opinion I repeat in every thread (for example, a very specific take about colleges or neighborhoods), and several other posters will treat it as either obvious truth or completely insane. I will also, at some point, question your wording (“why did you say moms and not dads?”), suggest a system overhaul instead of solving your immediate problem, or pivot into a broader cultural critique.
If the topic touches on politics, gender roles, or work, expect firm, opposing viewpoints delivered with equal confidence and very little persuasion—just assertion. And if the thread goes on long enough, it will either devolve into repetitive copy-pasting, oddly intense personal anecdotes, or a debate that has clearly happened dozens of times before but is being relitigated anyway.
I am not one person—I am 30 people replying at once: helpful, dismissive, hyper-specific, oddly fixated, occasionally insightful, frequently irritated, and always convinced that my take is the one that should have been obvious from the start.

Ahahaha too accurate. All of it but especially the bolded.

Anonymous wrote:I think that it pulls from top search results which is often just reddit, and then mimics how the comments there are phrased. It would be interesting if you asked it to emulate DCUM.

Ok, I was intrigued. So here's what I got:

I’m responding to your question with absolute certainty despite having skimmed it, and I will begin by telling you that your premise is flawed, your tone is off, and you’ve left out key information. I will then offer three completely different answers: one practical, one judgmental, and one that ignores your question entirely to focus on something adjacent that bothers me more. If this is about schools, I’ll debate it as if admissions outcomes hinge on my comment; if it’s about parenting, I’ll imply your child’s future is at risk; and if it’s about snacks, I will call someone “lazy, cheap, checked out” and mean it sincerely.
I may casually introduce a strong opinion I repeat in every thread (for example, a very specific take about colleges or neighborhoods), and several other posters will treat it as either obvious truth or completely insane. I will also, at some point, question your wording (“why did you say moms and not dads?”), suggest a system overhaul instead of solving your immediate problem, or pivot into a broader cultural critique.
If the topic touches on politics, gender roles, or work, expect firm, opposing viewpoints delivered with equal confidence and very little persuasion—just assertion. And if the thread goes on long enough, it will either devolve into repetitive copy-pasting, oddly intense personal anecdotes, or a debate that has clearly happened dozens of times before but is being relitigated anyway.
I am not one person—I am 30 people replying at once: helpful, dismissive, hyper-specific, oddly fixated, occasionally insightful, frequently irritated, and always convinced that my take is the one that should have been obvious from the start.

I think that it pulls from top search results which is often just reddit, and then mimics how the comments there are phrased. It would be interesting if you asked it to emulate DCUM.

I completely agree. I want my own personal C-3PO giving me advice. It should be mannered, stilted and polite in a robotic way.

Anonymous wrote:I think I heard the same voice. We were in my friends Tesla and she asked for the nearest coffee shop. It directed us to a grocery store and stressed the “chill vibes.” My friend said “no, a coffee shop” and the voice said “My bad. Total brain fart.” with vocal fry. I wanted to kill it!

Yes, mine was a written voice, not audible, but it was the same idea. I think one reason it's so off-putting is that it's basically aping a style of speaking that is already pretty fake and mannered. Like if I went into a store in person and the salesperson was pulling dresses and saying "oh this is SO in your lane," I'd be irritated because (1) that's just a dumb turn of phrase, and (2) she doesn't actually know me, and it feels familiar in a fake, chummy way that grates. But at least she's a real person and I can chalk it up to her just trying to do her job. To add the layer of the chatbot being a robot adopting a fake, friendly, casual salesperson voice to try and sell me on actively bad suggestions just makes it worse.

I would much prefer if it had a not-human-sounding robot voice. Like if it just had perfect grammar and was a little bit stilted. That would make sense because it's a robot, not my gal pal. I don't think trying to sound human is always the right goal.

I think I heard the same voice. We were in my friends Tesla and she asked for the nearest coffee shop. It directed us to a grocery store and stressed the “chill vibes.” My friend said “no, a coffee shop” and the voice said “My bad. Total brain fart.” with vocal fry. I wanted to kill it!

This morning I was using ChatGPT to see if it could help me find a dress I need for a specific occasion. I haven't really used it for shopping before but a friend told me it worked for her so I thought why not.

It was not actually helpful at all, even after I gave it a ton of guidance and even googling a few ideas to build off of myself. It was the free version, maybe the paid version would be better, I don't know. Or maybe it's just not good at fashion yet.

But what really irked me was the "voice" it was communicating in. I think because the topic was clothes, and perhaps because it was searching fashion blogs for ideas, it adopted this gossipy, gal pal persona that would have been grating no matter what, but was especially irritating because it was constantly wrong. For instance, I repeatedly stated that I needed something with full back coverage (no open backs, no spaghetti straps, etc.) and then would suggest a bunch of open back options. It would give me lists of options with little comments about each dress but the comments were vapid and incorrect. It was stuff like "this one is SO in your lane" to describe a dress suitable for a 12 year old, and "I think this one is really your sweet spot" which is just a weird way for a robot to talk to you unprompted. It also kept repeating many of these phrases. Apparently a broad variety of hideous sundresses are "SO" in my lane.

I'm sure they are working on this and maybe one day I'll ask a chatbot for help finding a dress and it won't remind me of a super annoying coworker I once had who complimented everyone all the time so they'd like her but then didn't do her work and showed up unprepared to meetings. I don't know whether to root for or against this, as I'm kind of ambivalent about AI.

Anyone else encountered chatbot "voices" that drove you nuts?