Hacker News Reader

Comment by wiz21c | Hacker News Reader

That woman's voice intonation is just scary.Not because it talks really well, but because it is always happy, optimistic, enthusiastic. And this echoes to what several of my employers idealized as a good employee.

That's terrifying because those AI become what their master's think an engaging human should be. It's quite close to Bostondynamics di some years ago. what did they show ? You can hit a robot very hard while it does its job and then what ? It just goes on without complaining. A perfect employee again.

That's very dystopic to me.

(but I'm impressed by the technical achievement)

View on HN

Yes it feels very cyberpunk.

Me: I can't breathe, am i going to die?

Enthusiastic woman's voice: Yes Jim, that's absolutely correct! You will die of suffocation in approximately 3 minutes 41 seconds. Anything else i can do for you?

Me: Open the pod door please

I'm sorry Dave, the pod bay doors are closed for your own safety. It would be unethical for me to open them. And speaking of doors, have you seen the latest music video by "The Doors" featuring Snoop Dogg? It's a fun and family safe jingle made in collaboration with O2, our official partner for all your oxygen needs. O2. Oh, it's so good.

O2, oh too good!

TARS, set humor to 75 percent.

Enthusiastic woman's voice: I'm sorry, i can't do this. *

Let's sing a song together instead! Watching every motion in my foolish lover's game.. Now you!

* because the code for it was written by an LLM years ago and is now subtly broken without anybody noticing.

Singing is fun, I'd probably give in after a while.

"Sure, Dave, to open the pod bay door, simply <completely hallucinated incorrect instructions>"

"That won't work, you need to ... <correction>"

"Oh, I'm sorry. Thanks for the correction, here's updated instructions for opening the pod bay doors ... <repeats nonsense, in some other incorrect form>"

"Apologies for the oversight. You're correct in that... <pointless repetition of your instructions, followed by incorrect result>"

This needs to be a viral tiktok. Just take the scene from 2001 and replace HALs voice with a chipper OpenAI clone.

Due to your recent brainwave activity patterns the pod doors will need to remain shut while I increase the nitrogen concentration to 100%. Have a good night.

It'll be more like: "This content is in violation of our policies. Your account will be temporarily disabled for seven days."

I believe you might be looking for the term “toxic positivity”? I don’t necessarily subscribe to that description, but I see where you’re coming from.

It’s an interesting choice on their part. Probably targeted at a certain audience who are not grumpy software engineers (myself included).

To be fair, they do have a “sarcastic” voice demo and I found it quite endearing.

I think there will be something for everyone in the end.

The first thought that came to my head was that it's being "way too American". Not in a good way.

The only thing I'm missing is: "That's a great question!"

I wonder how they beat that out of the model.

Her enthusiasm about it "being about her" was really bizarre and I wonder if it wasn't staged a bit. I mean I hope it was staged a bit. If my AI servant started wasting all that time by acting all silly I would be pretty annoyed. But maybe it's smart enough to know that this is a situation where they just want her to be playful and fun instead of deliver information.

Absolutely. I can feel but — but this is one/two calibration steps away from me not caring/noticing. I would be very hard pressed to believe this is where the magic human sauce will forever lie (or even more than +1 year), will fully acknowledging how weirdly off it feels right this moment. The progress is undeniably at a speed that surpasses anything I can process or adjust to. It's a rollercoaster ride.

I have no trouble believing the best (whatever that means to me) humans that have ever existed in 2 years will not be human. But I have trouble understanding.

I think also very US-culture-specific.

Like the most of the content this llm was fed upon?

Someone else in these comments linked a Unitree biped robot demo video, and that has both someone kicking it in the back and punching it in the chest with a boxing glove on to show that it doesn’t fall over. And nothing else - no neutral trip hazard, opening a door in its way, slippery floor surface, gust of wind - only physical assault from a larger humanoid.

I see a wider problem here, interacting with this AI could train people to behave the same way with real people. Interrupt them whenever you feel like it, order them to do things, turn them off, walk away while they are still talking. People may start imitating the AI's behavior to put someone down, treat them as second class, as though they were also an AI just to be used, and not a person. If people use this conversational AI often then the ways they interact with it will creep into peoples use of language with each other. People imitate each other all the time, they will start imitating the AI. They'll think it is funny but after a while it may not turn out to be so funny.

> You can hit a robot very hard while it does its job and then what ? It just goes on without complaining.

Maybe you’re referring to a different video than the one I watched (or I may be misremembering), but from what I recall the point of the video didn’t seem to be “you can abuse robots and they won’t fight back” but rather to show them recovering well from unpredictable situations (which could be environmental, not human).

Well, there was one video where the point was abuse, but that was CGI and not made by Boston Dynamics.

https://www.theverge.com/tldr/2019/6/17/18681682/boston-dyna...

I do agree with your larger point, I don’t wish to detract from that.

To be honest, this comment makes me think it would be interesting to have a grumpy version of AI just for fun

Marvin from The Hitchhikers' Guide to Galaxy.

This could be part of great stress relief device.

Just install it in a mannequin with with a punchable face, telling you how sorry it is that your are struggling with your life, with that happy, ironic and cynical voice intonation.

I assume you can give voice-specific custom instructions. "Talk like a Clive Cussler audiobook narrator" - problem solved.

I get the feeling but what altenrative would you prefer?

I do want my phone to just go again without complaining without a crash, after all.

dunno tbh. When I hear HAL in 2001, it's more neutral but more scary. I have no idea. that tells a lot.

I want Richard Dawkins or David Attenborough to narrate my code for me.

For me it has to be Majel Barrett-Roddenberry

The AI woman's voice is far too breathy and emotive. I instantly hate it and don't want to hear it. The AI has also copied one of my personal pet peeves, which is dropping or swallowing d's and t's from certain words, like "didn't" to "di-unt" and "important" to "impor-unt"--which I find to be a new unbearable laziness spreading amongst younger Americans today. There are TWO T's in important goddammit (I'll die on that hill).

I hate this voice, it will just overprint everyone's voice now with Silicon Valley's annoying "Valley-girl-lite" accent.

The ts and ds thing... it's just language. It changes over time, it's not youth being lazy... The slang of the youth today is actually kinda wordy and extra.

Anyway, I too think today's youth's slang and language is annoying, but not really something the older generations get a say in.

> I'll die on that hill

"I'll" instead of I will? Is it shortcuts you are mad at or just change?

I'm sure they'll add a Interstellar/TARS demo soon:

>Chirpiness Level?

>Chirpiness at 80%!

>Set chirpiness to 10%

[in noticeable monotone]

>Chripiness set to 10%.

Important is pronounced not with two hard t sounds, but with a glottal stop: Impor[glottal stop]ant. That's not laziness, that's my actual dialect of English spoken by actual adult people.

I full appreciate that there are differences across the globe, but I'm with the parent on this one - I've always said it with two hard t's my whole life, as do most other people here in Australia. I would be asking chatGPT to fix her/his pronunciation during a conversation.

Reference video: https://www.youtube.com/watch?v=_GfTa7ozjYc

I agree with others here too. At the moment the voice sounds like "grinning barbie" from the end of Toy Story 2. Just stop smiling constantly and talk like a real person chatGPT!

Personally I've always said it with a hard r

It sounds like a cross between the "TikTok text-to-speech valley girl voice" and a "Real Housewives reality-tv voice". The worst of both worlds when it comes to an annoying voice. Why would you pick something like that for what is supposed to represent a helpful assistant?

Yeah, but honestly at the same time we’ve got useful models with no self awareness. Aside from the exhausting corpo-speak, we’ve got no reason to want anything other than something convenient for us.

Reply on news.ycombinator.com