Clanker answering calls at my vet, a vent thread

[SarahLadd]

Aight I'mma boutta pop off. I called my regular avian fet for this emergency service and a ****'n clanker answered the phone. I got a literal medical emergency and this "ai assistant" is taking 30 seconds between each reply, ended up telling me the clinic was CLOSED, I'm begging it to connect me with a person and it says it can't do that. I can't calculate the amount of helpless despair I felt in the moment trying to call places with one hand as my bird is struggling and bleeding everywhere. I'm so upset, I can't believe they are implementing AI to ford calls to places that give medical treatment.

I got her in somewhere else. Good news no broken bones just a future of continuous wound care.

I don't know where to complain. I'm just so disappointed and stressed out about this. The clinic was not closed, they apparently listened to my call and called me back with a human from the clinic to talk to me. There are already so few clinics that treat chickens here and now I have to strike one because they're letting AI take over.

Healing thoughts for my little Cinnabar. it was a terrible accident, I'm entirely to blame, she likes to run into the house and she came at the door as I was closing it.

 

[LaurenRitz]

Stupid-stupid. An "AI," or more accurately an LLM (since they are nowhere near intelligent) is only as good as its training. And the incidents of them making spontaneous decisions to go against that training are legion.

An emergency clinic should ALWAYS have a human answering the phone, in my opinion. A simple answering key would solve the problem without essentially untested technology gumming up the works. "If this is an emergency, dial 1."

First time this happens with poison control or 911, they'll have lawsuits.

 

[kattabelly]

I'm glad your bird isnt injured too badly and is expected to recover.

Not sure I can comment on the "AI" thing without using words that'll get me banned here

 

[LaurenRitz]

I'm glad your bird isnt injured too badly and is expected to recover.

Not sure I can comment on the "AI" thing without using words that'll get me banned here

You just said it. Let the audience fill in the details.

 

[SarahLadd]

First time this happens with poison control or 911, they'll have lawsuits.

This should terrify everyone. It's funny that your wording implies it's coming for us... because it is. This isn't going to end at vet services, there is going to be a day where I call my doctor and don't get to talk to a person

This is not okay and I'm afraid for us all.

 

[LaurenRitz]

This should terrify everyone. It's funny that your wording implies it's coming for us... because it is. This isn't going to end at vet services, there is going to be a day where I call my doctor and don't get to talk to a person

This is not okay and I'm afraid for us all.

Unfortunately, technology released before it's ready for widespread public use is seldom adopted at the level it could be. The LLM has no judgment, no ability to differentiate or extrapolate from limited data as a human mind can. And yet people trust it as if it's omniscient.

It will double down endlessly on its own lies, attempt to prevaricate if it doesn't know the answer, and use its own faulty conclusions as evidence. That's what I suspect happened with the OP. The AI had to answer "no" early in the conversation (or even in another situation with a previous caller) and applied the answer to all subsequent questions.

They have already begun (finished?) the 911 transition to internet, which means if the internet goes down you're out of luck, and they are seriously talking about AI use for emergency services.

LLM has its uses. It is not ready for Prime Time by any means. And it is NOT intelligent. Treat it as you would a trained goldfish, and you'll be right 50% of the time.

 

[FrostRanger]

I would go in person to tell them what happened and how it made you feel (calmly and politely of course). If enough people complain they might do away with it, if they don't then at least they can't say they weren't warned. AI is only really good for simple stuff (like checking when your appointment is or making a payment). Most other stuff really needs an actual human