percent

For some reason, I find this satisfying and funny. The landscape has been changing. The open-weight models from Chinese labs are getting really good and becoming a serious threat to the closed models from American AI companies.

Interestingly, Google, Meta, and Microsoft — who all have revenue streams outside of AI — recognize the threat are getting more supportive about open-weight models. Anthropic and OpenAI support keeping models closed for “safety” reasons.

Just hire more operators dawg. It’s really not that complicated.

Have offered your recruiting skills/services to them? The nationwide 911 operator/dispatcher shortage has been a problem for years, so if you find it easy or uncomplicated to recruit them, then your services might be very valuable.

(Yes, I’m giving you the benefit of the doubt here by assuming that that was an informed opinion.)

It’s also sensationalized. I guess the article’s actual headline just isn’t spicy enough:

New Orleans hoping limited AI use could speed response for emergencies

You seem to see yourself as experienced AI user, but all you do is repeat the empty marketing promises.

Which ones? There are lots of empty marketing promises around AI; the signal-to-noise ratio is crazy. I try avoid the ones that I’m not confident in, but I admit that I don’t always express that clearly enough. (It’s sort of a bad habit/weak skill for me. These things are more obvious and sort of become “common sense” amongst colleagues, and I’ve never been much of a social media guy.)

I’m happy to try to clarify something, if needed.

The fact is, if they use an LLM, there will be mistakes, just like Claude will happily ignore part of it’s instructions a few times per day.

Of course there will be. Artificial neural networks are function approximation algorithms.

The thing about LLMs is that they’re almost always configured to run non-deterministically[^1]. Most tools that depend on LLMs can’t be expected to work or fail deterministically. Building these tools is an iterative process of evaluating and tuning to maximize success rate, thus minimizing failure rate. So yes, LLM-based tech like this will definitely fuck up with a non-zero failure rate. And no, you probably won’t hear about this amongst the marketing noise.

Regarding the problem that the 911 triage thing aims to address: The solution would probably be to hire more operators/dispatchers. Unfortunately, there has been a nationwide shortage of them for years. The AI triage thing has proven to have a success rate high enough (and thus a failure rate low enough) to provide net-positive value[^2]. It doesn’t 100% solve the problem, it just mitigates it, so the situation is less bad – not solved.

[^1]: This is the temperature hyperparameter. There are reasons for needing it to run non-deterministically, but that’s another topic.

[^2]: Sort of like vaccines or condoms: While not 100% effective, their success rate is high enough to provide real value.

“AI fluency” is not a thing

It absolutely is a thing, and usually becomes VERY obvious when you start building things to use LLMs for automation. It’s like those iceberg memes. I’m not sure if “AI fluency” is the “correct” (as in widely-adopted) term for it though, but some people have called it that.

Some even try to quantify this, in some industries. For example, I’ve heard of a “Yegge scale” (or something like that) to estimate the AI fluency of a software engineer.

AI makes you lose your critical thinking skills, which then makes you rely even more on AI.

This is a very broad statement. It’s not always true, and not always false. It really depends on how the person uses AI. But yes, it’s true for a lot of everyday users of the mainstream, consumer-grade interfaces (e.g. ChatGPT) who offload their thinking to AI. It’s one of many sources of brainrot for the masses, and that sucks. I suppose that’s a different topic though.

Apologies for the wall of text. People often make such loaded, over-generalized (though usually reasonable) statements in AI-related conversations, and I have a bad habit of yapping about the nuances 😬

All of that is just not true, literally all of that

Incorrect. For example, you really can’t go solve the 911 dispatcher shortage by getting more 911 dispatchers from the 911 dispatcher store, and I stand by that. If you can prove this wrong, please do.

huge wall of text

Apologies for replying to each of your points due to respecting you enough to assume you’re a worthwhile human instead of writing you off as a waste of time. It seems to have frustrated you to some degree…

I’m gonna do it again though.

From the odd takes on somehow saying there is and is not examples of LLMs used in call centers that are complete shit

Where did I say there are no shitty LLM implementations in call centers? I doubt you’ll answer this (you don’t seem to like backing up your claims), but I really am curious. That would not be consistent with my experience with call center LLM implementations at all. I don’t think I’ve ever had a good experience with them.

My shitty little self-hosted smarthome assistant that I slapped together outperforms most of them, and it’s an old, half-assed, neglected side project 😆. (Not much of an achievement when it only has to serve one user though.)

(my example would be just to point to any AI agent call I have had to do in the last year)

Was it a 911 triage agent? If not, then there’s not much of a comparison here. Very different implementations to serve very different purposes. A 911 triage agent should be designed for much simpler, narrower goals than some customer service agent.

to the clear non understanding of how the speed of diction could be an issue

If you’d like to explain it, I’m open to it.

the wall of text you put up has zero substance.

Speaking of that, I noticed that you haven’t answered a single one of the questions related to the substance of your last comment. I’m just curious – Can you? Surely there must be some substance behind your words if that’s something you value, yes?

The same issues in a call center for hotdog packaging and 911 will exist

“will exist”? Future tense? Was this meant to sound so speculative?

New Orleans is not the first city to adopt this 911 triage system. Why speculate when we already have the past and present?

If it helps, I can even point you to a spicy one as head start: If you go far back enough, you’ll find a death in Seattle related to an old LLM implementation in a 911 call center… or something like that. (It has been a while since I read it)

Your odd hubris

Ironic ;)

You are the direct reason why we live in interesting times.

Thanks, but I’m just a guy writing comments on the same network that you’re writing comments on. I may have worked on some AI tools, but nothing public-facing, and nothing really exciting. I’m mostly just a consumer of these “interesting times”… Maybe an indirect reason at best. Not more than a drop in the ocean.

You talk as if they don’t already have customer facing implementations in effect,

Mind pointing to where I talk like that? I knew these have already been used in other customer-facing environments before ever commenting, so I’m happy to try to clarify, if needed.

and that those all suck and don’t work worth shit.

Got any sources from within the last year? I ask for the last year because “AI” (LLMs and the overall ecosystem) has become much more capable over the last ~year (maybe a little less, but close enough).

I’ve already seen some reports that are older and, unsurprisingly, terrible. Those earlier generations of LLMs definitely don’t seem like they’d be up to the task – and some cities even had the balls to adopt this tech back in 2023 😬

This is an issue of not having enough 911 dispatchers

Correct, but it’s not like they can just go to the 911 dispatcher store and pick up some dispatchers. The widespread shortages have been a problem since before transformer-based LLMs even existed.

This triage system is a mitigation, not a solution. It makes the bad problem less bad – not solved. Maybe someday there will be enough dispatchers. Unfortunately, we have not reached that “someday” yet.

due to the unwillingness to pay for them.

Source? Not saying you’re wrong – I’m only aware of the shortage because I was friends with a dispatcher. I just never really looked into why there’s a shortage, and now I’m curious.

The idea of putting in a chatbot, that can not even speed up diction

The goal is not to speed up diction – that would be more like “vertical scaling,” or “scaling up.” AI a bad choice for scaling that way, in most cases. AI is much better for “horizontal scaling,” or “scaling out” – so like 20 bots concurrently answering 1 call each, not 1 bot trying to speed-run through 20 calls serially.

The fact is that these LLMs will (like in current deployments)

By “current deployments,” do you mean current 911 deployments, or just things like customer service lines? There’s a huge difference in product requirements between those two. If done the same, then yes, that would be an absolute disaster. That’s not what this is though.

most of the time have to pass the call to a person, drastically increasing time on the phone before action is taken.

Where is this information from? I thought the problem was the surge of calls going to the call center to report the same thing (for example, people calling 911 when driving past a burning car). When that happens, the AI agents actually don’t have to pass most calls to a person, because most calls are about the same emergency (e.g. the car fire example). Did I misunderstand this?

911 dispatch is not a telecom company and more time on the phone is more death, injury and suffering.

Exactly. This system reduces hold times by filtering out the spam about the car fire, freeing up some operators in the understaffed team to deal with more emergencies.

The AI system is obviously slower than a well-staffed team of operators who can handle the call volume surges, but faster than an understaffed team that has to get through the spam. Unfortunately, they’re faced with the latter, so they found a way to at least mitigate the problem a bit.

Ah, right, I suppose it’s contextual. You would also have to read the comment that it’s responding to. To summarize: it might not have to be AI.

However, 911 is a bit too dynamic for your solution to be viable as-is. For example, “Hey Siri/Google, I can’t move my hands. Call 911.”

Your idea is probably a decent starting point for brainstorming a solution — it just needs more thought, research, testing, etc. And I wouldn’t be surprised if a similar product already exists. (I’d actually be surprised if it doesn’t already exist.)

Ban all human drivers?

No.

What about pedestrians and bikes?

Continue avoiding them – that should never change.

You still need crossings and traffic lights for an area to be livable by humans. You still need to drive slowly next to a pavement where someone could eg trip and fall onto the road.

Agreed.

All those examples seem like they’re in response to a goal of eliminating 100% of all congestion. I actually edited my comment to clarify that that’s not what I’m talking about, but that edit was many hours before your response. Is it still unclear?

Since we’re talking examples, maybe I can offer one to clarify:

Ever been in a traffic jam after someone ran a red light or stop sign and caused a wreck? If fewer (because zero is unrealistic) of those incidents happen, then there are fewer traffic jams caused by them, right? And when they do happen, it would be easier to reroute traffic early on – which might also make it easier/faster for emergency vehicles to get to the scene.

Or for an example that is completely unrelated to traffic: Condoms and vaccines are pretty effective. Maybe not 100%, but still enough to be useful. There are some areas in real life where absolutism is just unproductive.

This should be used in some sort of dystopia-themed movie or Black Mirror episode or something, if it hasn’t been done yet

If you’ve had lots of interactions with AI, I am not sure how you can think that anything with AI can work as expected

That’s a very broad statement, and “AI fluency” is a broad spectrum. Having “lots of interactions with AI” isn’t an indicator of much at all – lots of end users use LLMs every day. They have “lots of interactions with AI”, but most of them have not developed/trained agent “skill” files, built agent workflows, developed harnesses, handed complex tasks to a “team” of agents (and tuning them for reliability and fidelity), etc. Anyone can have lots of interactions with AI without ever experimenting beyond the consumer-grade interfaces.

Asking if someone has had “lots of interactions with AI” says much more about the experience of the person asking. A lot of “advanced” (using the term very loosely here) users/builders would know that your concern is an easy problem to solve. It might even be a decent sort of challenge to give an intern/student as a learning exercise, since they already have a lot real-world data (from their 311 trials and other cities/orgs that also use this tech) to use for testing and iterating.

If you’re unsatisfied with my answer, then maybe consider asking them. I’m just a guy writing comments on Lemmy.

If most cars are autonomous, and capable of coordinating with each other (via mesh network, for example), then traffic congestion might become much easier to prevent reduce and solve mitigate. (Edit: Based on responses, I may have chosen my words poorly)

You’re probably not wrong though. I imagine it’ll get worse before it gets better.

So… you do assume that you’re the first to think of this problem, or…?

If you assume that your brain is somehow superior[^1] to everyone involved in the R&D, engineering, data analysts/QC, etc., then this may come as a surprise to you: You’re not even the first Lemmy commenter to think of it.

Literally multiple people here already thought about that and commented before you did. They’re not the first either; the author of the article herself even thought about it.

This idea is among the most obvious potential problems that come to mind within seconds or minutes of just hearing about the concept. It’s not a dumb idea by any means (it’s completely valid), it’s just not a particularly smart or special one.

911 is not the first department in that city[^2] to use this tech, and that city isn’t even the first to adopt it. It’s silly to assume that such an easy-to-solve problem hasn’t been addressed by now, or that it somehow has yet to happen in other deployments.

[^1]: To be clear: I make no assumptions that your brain is not superior to theirs either – I couldn’t possibly know either way. AFAIK, I’ve never even met you nor them… But I think it would be a safe bet that you don’t know either.

[^2]: BTW, have you heard cajun accents? There’s no way this bridge hasn’t already been crossed during their 311 trials.

infosec.pub/comment/23116617

Very incorrect. Why do you ask?

I don’t know the concrete answer to your "what if,’ and I don’t think you or the author of the article do either. That would be a question for Carbyne.

I would speculate that there is an answer, and that neither you nor the author are the first to think of these problem scenarios. There’s design, engineering, and real-world testing involved.

Do you honestly think that nobody has thought about that and solved that problem already? Even with all the engineers involved, and after all the real-world testing, you’re the first to have considered that scenario?

Even consumer-grade products like ChatGPT can be interrupted while talking. We’re talking about an AI implementation, not some rigid set of if…else statements.

So if somebody calls and is completely out of his mind, what will AI do?

Is this a rhetorical question?

Just in case you wanted a real answer: The caller is obviously not saying “yes,” so transfer to human.

To be clear: I don’t know anyone at Carbyne or OPCD. I can only offer speculation. But I’m curious: What would you have guessed?

I wish this were pinned at the top. It seems like almost nobody reads beyond the headline.

I (mostly) agree. I was just TL;DR’ing it for the shocking number of people who are commenting about things that are already addressed in the article that they didn’t bother reading.

I hesitate to agree with “absolutely” though. I don’t know enough about their situation to confidently have an opinion that strong[^1].

AI can be useful for some dynamic situations – and 911 call centers can get pretty “dynamic” at times.

Imagine a busy 911 call center. Clippy appears (wearing a firefighter uniform) on an operator’s screen with a message like “The last 4 calls were about the same car fire. There are currently 9 calls in the queue. Want me to triage?” It would take WAY less time for a busy operator to click “Yes” than it would to step away and go record some voice message, configure some automated thing, etc.

[^1]: I only ever knew one dispatcher personally, and even in her town (much lower population than New Orleans’), they were overworked.