Anonymous #1 6w

Also the fact that this is just a basic GPT while the 911 line specific AI would obviously be much more tailored and tested for edge cases

-3

Anonymous #2 6w

I don't think that this idea is without flaws but human 911 operators are already fucking horrible so if we can actually use this to improve the system then it seems great

-12

Anonymous #2 6w

Have you ever actually talked to a bot on the phone and tried to get a question answered?

49

Anonymous #3 6w

Yes but these are not generative AI bots, they're the same voice to text algorithms as in the 90's.

-15

Anonymous #3 6w

Yes it took me a second to realize it was a bot but the cadence was a bit uncanny and there was fake background audio to make it sound busy. But I scheduled my derm appointment with no issues.

-4

Anonymous #2 6w

You might be the only other person on here with a brain bro

-10

Anonymous #2 6w

Some places do have LLM secretary bots and they’re pretty good ngl.

-12

Anonymous discotheque 6w

Actually thats true, I wasn't thinking of it off the top of my head (because my mind goes to the piece of shit dial up style AIs instead of LLMs) but now that you mention it, my apartment complex does this and they're really helpful, like the bot can tell me if any packages have arrived for me in an instant.

9

Anonymous #1 6w

You realize though this is only training off of stuff that has already happened, right? Like the whole “ordering pizza” thing is semi well known at this point. What happens if they get a call where someone is talking in code about something not even related to food in an attempt to get the message across that they are in danger? The Ai isn’t trained for that at all, and it can’t be, because it’s a new and unknown situation

26

Anonymous discotheque 6w

Yeah last time I tried talking to one of those on the phone it transferred me between four different “departments” until it realized only a human could help me and told me to call back later

12

Anonymous #6 6w

It would still ask “are you in any danger?” And then would go down the same tree

-5

Anonymous discotheque 6w

You got a source for that one buddy or are you just assuming the AI would say that?

15

Anonymous #6 6w

The AI has access to over a trillion bytes of information and can easily decipher any code you give it. A human probably not. Even obscure references AI will know. I don’t think you understand artificial intelligence and LLM.

2

Anonymous #6 6w

What if I start talking on 911 in Somali? Is your average human 911 operator going to understand it immediately? The AI will.

-1

Anonymous #2 6w

DING DING DING

1

Anonymous #1 6w

Again, it’s going off of what is already preexisting. That trillion bytes of information is already existing information. A new and unique scenario is unable to be addressed, because the Ai has quite literally not been trained off of it (unless it has some way of knowing the future ig, in which case we got an even larger problem)

8

Anonymous #6 6w

Brother there is no new or unique situation. Everything that’s happening now has happened. Only thing is if aliens landed on earth. And if the ai cannot recognize the pattern I highly doubt a human will know what to do

-1

Anonymous #6 6w

Again you don’t understand how AI works. Do some research then come back here

-1

Anonymous #2 6w

There’s already services that exist where 911 call center employees can start a three way call with translators who are specifically trained for such a scenario, but nice try

3

Anonymous #6 6w

There is no fucking chance any 911 operator is going into this much detail to code break 😂 You are literally making up scenarios to get mad at

-2

Anonymous #2 6w

I've seen videos of 911 operators hanging up on life or death calls because they get bored or annoyed.

-1

Anonymous #2 6w

People in dangerous situations famously have the time to start a book cypher call

4

Anonymous #6 6w

I know it would be able to say that. All you need is a system prompt and it will do it. It is literally out here writing new proofs for the Riemann Hypothesis today. Seems pretty evident it could handle it.

-1

Anonymous #2 6w

Cool and there’s been suicidal people who took their lives because Ai told them to

5

Anonymous #6 6w

Honestly if you're in a situation that requires you to be talking in such a complex code that an AI can't figure it out, just call and stay silent. They will send people eventually.

0

Anonymous discotheque 6w

Got it so you just made it up because you “know it would be able to”. Wonderful

11

Anonymous #6 6w

And there’s people alive today because AI told them not to and told them to contact the suicide hotline

5

Anonymous #6 6w

I'd bet money that they just got AI to rationalize their own thoughts of self harm. This also isn't a suicide hotline, its an emergency services hotline.

-1

Anonymous #6 6w

And AI didn’t explicitly tell them to do that btw. Read the case

1

Anonymous #6 6w

Like I said all it takes is a system prompt to ask “are you in danger”

-1

Anonymous #2 6w

100% what happened

1

Anonymous #2 6w

Suicide hotlines famously don’t even interact with police. That like never happens, they’re two entirely different entities

1

Anonymous #6 6w

https://www.kuow.org/health/2019-03-06/she-called-911-and-said-she-was-suicidal-then-she-was-charged-with-a-crime Also this brings into question, would you rather have an imperfect AI or a system that charges suicidal people for trying to contact help?

-1

Anonymous #2 6w

Right now, since human 911 operators are a limited resource, we actually do charge people for calling 911 with suicidal ideation.

-1

Anonymous #2 6w

I’m sorry, do people at call centers charge people with crimes or is that police officers?

7

Anonymous #6 6w

Neither.

0

Anonymous #2 6w

Ok nice to see you have at least one foot in reality

7

Anonymous #6 6w

Do you not recognize that the misuse of 911 charge only exists due to the limited quantity of 911 operators?

0

Anonymous #2 6w

I do. The solution is not Ai. In the case of operators, one of the main reasons for staffing issues is that lots are burnt out and stressed having to deal with a lot of stuff. This workload could be massively lessened if we had dedicated mental health professionals who respond to calls and make that a separate department and hotline. Additionally, part of the reason for the burnout is low pay. Lets fund these centers more and give them a much needed pay raise

1

Anonymous #6 6w

Not the point, a human operator would have to take the time to connect to a translator service while the AI could just talk in that language (or literally any) from the jump.

0

Anonymous #6 6w

So if an AI ever makes any mistake, it can be blamed completely on the idea of AI, but when humans do the same, it's just a result of external factors out of their control? 😂

0

Anonymous #2 6w

So we’ve jumped from staffing shortages to instead talking about who is to blame when a call goes wrong. Just keep gish galloping more dude lmao

4

Anonymous #6 6w

Bunch of cope bro, it's not a gish gallop at all. You are talking about how people are not inherently flawed for use as 911 operators despite not being available enough, while saying the AI is inherently flawed in use as 911 operator for similarly easily fixed flaws

-1

Anonymous #8 6w

Because you have the basic model. My model instantly understood the assignment. I can even tell it I’m in danger in Morse code and it’d know

0

Anonymous #1 6w

For the ones that say we need comfort too

0

Anonymous #1 6w

A human operator would not be this transparent with you

5

Anonymous #1 6w

Holy shit you’re having a meltdown over AI. Why would you ask a taco place to comfort you?? And “I’m scare too… 🥺🙏🏻” go outside dawg. Bask in the sun or something

2

Anonymous #1 6w

Also a trillion bytes of data is only 1TB. An AI that can fit on my Xbox external hard drive seems like a really shitty AI.

5

Anonymous OP 6w

I only included that because someone in the original thread was tryna make a case that AI can’t comfort you lol

4

Anonymous OP 6w

1 trillion terabytes of data

1