Hacker News Reader

Comment by Molitor5901 | Hacker News Reader

I've seen it affecting in a few ways: Less customer service employees at call centers, replaced by lengthy, egregiously pedantic AI systems. Some fast food places are replacing employees with an AI drive through. I've also seen AI health coaching services that removes the dietitian/nutritionist/health coach from the equation. Not quite AI, but getting there, are the self checkouts slowly replacing front line cashiers, etc.

So I think it is affecting the job market, but not in the white collar, higher paying jobs that people tend to notice.

View on HN

What is AI providing in the drive through situation? It just feels to me like much of this is "AI" in the business buzz sense, not generative LLMs or whatever doing any kind of work. Like, we have all been struggling with call center "AI" for a long time before this, I personally have not experienced an LLM chatbot call, but plenty of things asking me "is this information correct?"

Like how would some of this even work in reality? Can you go through those drive throughs and ask it to recite a sonnet about chicken nuggets? Clearly no, but then it begs the question of what the idea, the purported advance, even is here with much of this. Like we have had relatively advanced speech recognition for a while, I don't see the added utility or need of being able to go through the drive through and saying: "the number of hotdogs I want is a prime number that is more than 2 but less than 5."

It just feels so clearly silly if you stop and think about it for two seconds. So many hammers, not enough nails... We are just banging at walls at this point.

“I’d like 2 cheeseburgers, and 4 fries. No mayo or mustard. Actually make one of them a double, and one with bacon. Oh how much if I make the first a combo?”

You think this conversation could be handled with the tech of 4 years ago? Siri can’t even turn off the lights and tell me a joke in the same request. Humans do not deliver all information in order (eg. The all the instructions refer to the burgers not the fries, but you only know that because you understand the essential nature of fries and what they typically include). That’s what AI in the drive thru is for.

I'm not sure current tech could reliably take that order, honestly. There's essentially 0 chance it would try to disambiguate the meaning of "one of them", and from there it's a tossup whether you'll get a double cheeseburger, a double box of fries, or double mayo.

It can't. Not reliably. I think every major chain that was trying it has ripped it out.

It'll definitely be a thing within 5 years, max, but it's not mature enough for production yet

Current tech is pretty dang close. I gave the order to ChatGPT and it parsed it almost perfectly [0], even handling the ambiguity about what happens if you add a combo to an order that already includes several fries à la carte. The only thing it missed is that I didn't actually order the combo (but merely want to know how much the upgrade is), but I'm sure some fine-tuning could solve that. (Come to think of it, a fast food restaurant would consider this implicit upsell as a feature.)

The main challenge AI would face is people who come by at 3 AM drunk and stoned, indecisively slurring through their order, but I imagine there'd be a system to redirect these edge cases to an actual human.

[0] https://chatgpt.com/share/68ba2233-9f48-8011-905a-c69cc5e91b...

Pretty dang close isn't the same as accurate for an exchange of time and money. Voice->text, with a noisy background, is a particularly hard problem. Especially with hardware not designed to limit background noise. Try it. Whisper is still the leading speech->text model in our tests, but add noise reduction, echo, diarization, etc. It's a hard problem.

>Come to think of it, a fast food restaurant would consider this implicit upsell as a feature.

Yeah, just what every restaurant manager wants: to deal with customers who paid more for things they didn't order.

To be fair, this would trip up a lot of _humans_ as well.

I agree with sibling replies but more tangentially maybe: why is it that sometimes the point of these things is that I do not have to modify my behavior at all, while the restaurant can pay one less person, but other times the point is all about modifying it so the company can pay one less person?

Like here: if the restaurant really wants to get rid of their intercom person, why not make it self checkout, no AI required? What is actually saved or gained either way? There is nothing intrinsic about this situation that requires me to use natural language to order something. People order tons of food online these days anyway!

Like I just dont think it makes sense and I also probably don't think the economics of this would work out with fast food restaurant scale.

Again, just step back and think about it for a moment: lots of this really doesn't make sense. The world is not really full of tasks a good prompt can solve. There a million things that aren't "produce this python script" or "summarize this article probably correctly."

Why can't it just be what it is? Why does it absolutely have to be everything or nothing? So much of the thought around this feels so clearly wrong headed, its just starting to feel truly absurd.

I stopped at chicken restaurant in Georgia, and the drive through speaker was clearly AI. It sounded like a computer. It got the order wrong when it repeated, and stopped responding. So with some.. trepidation I said "Hello?" and after a few seconds an actual human (with noise in the background) responded, took my order, and all was right.

If they can apply AI to customer service phone queue hell, why not to most customer service transactions - my thinking goes - what will be next? Hospital? 000? Tax office?

It's a horrible tendence. The first thing I ask to such AI customer service chats is to let me speak with a real human.

I hate AI CS with a passion. It has neither the information or the authority to be effective.

I've known so many people who were destroyed by customer service work in call centres though. Most turned to drugs, some to suicide, the lucky few made it out but still carry the trauma. Timed toilet breaks, eight hours of angry customers, for basically minimum wage with no benefits. It's one of the worst environments for a person to be in and seeing it become automated I'm definitely aware of there being upsides as well.

That's only an upside if there are better alternatives for those people. If there are only worse alternatives, then obviously it's worse

I've worked in call centres a couple of times and found it depressing, but I've also worked in a restaurant washing dishes on sixteen hour shifts. To be honest, call centre was better

Where is the line of what you find acceptable then? As long as there's no a alternative anything is ok?

Like I get what you're saying but some work is just cruel.

Yes, and some work is crueller than others.

What I'm saying is there's no upside to just getting rid of an option

I outlined the upsides.

At least in Canada you can legally demand what it promises you.

It's funny because I had my first encounter with one last week and my first reaction when I realised it was a robot was "let me talk to a human".

I am mostly ok talking to a bot for FAQ still queries but I just don't want to interact with a machine when things are going wrong. I want someone with actual empathy even if they refuse to use it.

Empathy is probably part of it, but I guess what matters to me is someone able to think outside of the box at least a little, and try to solve an actual problem - since the problem is likely caused by the shape of the box.

The biggest issue with these systems is they are designed to handle the common case. But if I had a common case, I literally wouldn’t be having an issue!

Humans usually recognize it pretty quickly if you explain it to them, ‘AI’ usually just keeps steering you to the same box.

Though recently I did hit a system that immediately sent me to a human when I described the problem, which was refreshing!

I guess they are made for the common case since a lot of people's first actions is to talk to someone and most cases are common. Since I usually have already gotten through the FAQ, I also rarely am helped by the AI CS.

Let me translate this to Senior Vice President of GenAGI innovation speak:

"omg our AI-interaction KPIs went up so much!!! Customers love to chat with the AIs, they send messages back and forth really rapidly like 10 times in the first 30 seconds!!" (to get to the humans...)

time to lay off everyone!

And then everybody clapped and cheered

Walgreens is really pushing the AI phone system these days. It lies and says it can handle most requests, which I make and it tries to pawn me off on the website. The website which I had already been to, which recommended I speak with a pharmacist. I swear by the time it’s passed my call through it sounds like exasperated when it says “Just so you know, the pharmacy is open. . . .”

You have to say “speak to the pharmacist” and it reminds you it can help so you say it again and it goes through.

I’ll try that next. The flow use to be “pharmacy” “something else” phone answered by pharmacist. Then they threw Pharmacy Technicians into the mix. Then they centralized the pharmacy technicians into a call center.

I hate having to call. I hate having the system be insistent on hearing my problem first. 25 years with the same medications, if I’m calling it’s for something that is not usual. And if the system were truly trying to be helpful it would realize I’m calling about the insulin that is delayed when I specifically requested pickup today, and maybe it could figure out I need to transfer it to a pharmacy that has it in stock.

I just say "speak to a pharmacist", wait until it starts to say something about how it can help and then again "speak to a pharmacist". I only get a controlled medication though so maybe it can be used for others

> Not quite AI, but getting there, are the self checkouts slowly replacing front line cashiers, etc.

There's some AI involved at some retailers - I bought 2 identical items and the second wouldn't scan at the self checkout, so I grabbed the first item and scanned it again, and the camera-watching, object-detection system threw a fit (and played back the video of me). I had to call a human to complete my purchase. My suspicion is it is smart enough to detect that I moved an "unscanned" item from my basket item into the bagging area, but not smart enough to figure out I wasn't trying to cheat.

That sounds dystopic and I'm not sure it'd even be legal where I live.

Here they simply have a + button so you can set the amount of the item. No need to scan all of them.

The problem with the + button, is a user has two yoghurts, so scans and hits the + as they're both $2. The problem is that one was blueberry and once was strawberry.

Forcing users to scan everything fixes that but at least.

Making the users take longer isn't a concern of the shop.

But those items aren’t identical and should be scanned separately.

I’m in Portugal at the moment and went to their version of Target using Sensei [1] technology. I was with my little one who was grabbing things off the shelves while I put them back, and I was using a stroller not a store cart. The system worked flawlessly including consumed cafe purchases (coffee, pastry, and bread) along with per kg fruits and vegetables. One of the odd benefits is that I felt a bit safer with my little one roaming freely as I knew he was being carefully watched.

[1] https://www.sensei.tech/

Those cameras are not there for the benefit of your toddler.

How can some big corp. recording and processing what he does possibly make you feel a bit safer?

These systems are often far more stupid than you think. A lot of them just block themselves if they think your hands went somewhere they shouldn't be and defer to a human then.

Wegmans recently introduced carts with built-in scanners and scales to allow you to scan as you shop—I really like them overall, and use them every time I go there (which is only once every few months; it's 45 minutes away), even though I often need to get a second, regular cart to offload some of my bags to partway through shopping.

The one problem I've had with them is that they have a tendency to get confused if you try to scan more than one of the same thing in a row, and occasionally I'll have to go through quite a bit of trouble to make sure that I do pay for the 3rd item, and not just the first two...

Do you shop in Sainsburys in the UK by any chance?

This has happened to me too.

Yes, but when Marc Benioff says he laid off thousands of customer service agents, the reporting is "Salesforce cuts tech workers using AI." The narrative in media is a total mess right now, and there are many in VC and AI-related companies ready to help muddy the waters further for their own benefit. Obviously, many small companies follow the media narrative.

I work in the media space. AI is absolutely ripping through film, TV, and advertising.

Several medium sized studios I've talked with are bidding $50k for projects (eg. Netflix, HBO, Proctor & Gamble are typical clients) they used to bid $400k on, and they're winning more contracts. They don't need to shoot in person in Venice for pharma ads or animate elaborate TV show intros anymore.

This is having a huge impact to the fundamentals of how they do business. They haven't laid anyone off yet, but they're talking about the ramifications if this gets cheaper.

Have these studios actually shipped any of their projects to their customers yet? Do we have feedback from a quality perspective?

It's quite easy to promise dirt cheap services and get paperwork signed.

> Do we have feedback from a quality perspective?

... Have you watched YouTube (without attempting first-party ad blocking) recently? The ads created with AI are pretty obvious, and pretty bad.

My local news has started using AI bullshit they would've used B roll for in the past. And it's obvious. And it's very jarring.

Many people think all plastic surgery looks bad because, by definition, you don't notice the good examples.

Even if every ad on YouTube were AI-generated now, there would be enough bad examples for me to be negative on the entire idea.

You're not the customer.

Also, if the viewer doesn't recognize or care, then it's a moot point.

There is tons of bad CGI, but that hasn't stopped its near universal adoption.

Lol CGI makes up a tiny portion purely because economically it makes sense.

Yes. You wouldn't recognize them as AI.

These studios are doing a lot of roto and comp work. It's highly touched up and edited.

What component of the production process is the AI being used for? Is AI video now good enough for green screen backgrounds or something like that?

I can't speak for the industry side but as a consumer, I've noticed many cable TV ads in hotel rooms now are clearly using AI generated video. It looks like shit.

This is going to be the "bad chromakey" of this particular time period in terms of weirdly prolific visuals in media. Or if you prefer, the ads you used to see on late-night TV that were clearly broadcast from a poor quality VHS.

Cheap bullshit has always hung around our media apparatus, and it's just that: cheap bullshit. Tbh I just note it in the same way I've always done: well, that's a company I'm going to avoid doing business with if at all possible.

Now you mention it, I can see there's a lot of demand for very cheap video ads on YouTube and such and I can see those kind of productions using AI slop and not really caring. I was just surprised by the calibre of client the poster above mentioned, such as HBO and Netflix and such. That sounded more like AI video making an impact in higher class professional work.

I mean I'm sure they would LIKE to use AI. What sane company wouldn't explore the possibility? That said I think any serious creative team is going to run into headaches with it really, really quickly and give up on it.

They do. I watched a person quit because they had to hand edit video frames that came out of AI. It would have been cheaper to do traditional VFX.

It's very powerful for various parts of VFX workflows. It's not gonna just be a full prompted shot, but more a means of creating and manipulating smaller elements in a shot with much less manual labor than before.

What are they doing with AI instead of eg. shooting in person in Venice?

>Several medium sized studios […] don't need to animate elaborate TV show intros anymore.

What projects are these studios doing for HBO? Its shows generally have high enough production value that AI slop in intros would be a no-no (unless this has dramatically changed under Zaslav's leadership).

I believe you, but what I don't understand is why aren't these companies producing custom models tailored to their in-house needs? You'd think that Hollywood would have the most advanced image and video generation models, after all the studios have high quality training data and more importantly, they have the IP rights to that data.

Meanwhile back in reality it's Google that is massively ahead of literally everyone.

Self checkout in supermarkets apparently is quite a rocky thing, some chains see profits go down after implementing them. The potential for catastrophic mistakes in LLMs is very high, like ordering 3000 big macs.

Sure in the case of 3000 big macs it would be caught by the buyer or the cooks, but ordering 6 instead of 3 will not. This will cause complaints, complaints need more people to handle, etc.

Honestly those problems will get sorted

But I wonder what the effect will be like otherwise, is it gonna be a turnoff knowing you're talking to a robot?

Like what if your cafe barista's were replaced by robots?

Maybe I wouldn't mind so much at first but I'd probably just switch to some place where there were baristas, because why wouldn't I just get a canned coffee from a supermarket if "cheap" was all I cared about

There was a fun video floating around a few days ago where someone tried ordering 18,000 cups of water... It certainly got them a human being lol

I once called a hotel. A suspiciously regular voice answered. I asked if it was a machine. It said yes, it's [AI marketing bullshit here]. I asked if it had a room. It said no, because it's [AI BS] and doesn't need a physical presence such as a room. I hung up and called another hotel.

That was a major success, according to the KPI the customer inquiry was resolved in under 60 seconds all with AGI!

This could have been an amazing short sci-fi story.

A lot of the applications listed don't need humans and probably don't even need "AI" either. At least not beyond speech to text.

When my Xbox 360 hit the red ring of death, I called in to Microsoft support and went through the flow to replace it with just pre-recorded responses and speech to text. This was in 2007.

> Less customer service employees at call centers, replaced by lengthy, egregiously pedantic AI systems.

Is this actually worse than being on hold forever to talk to someone following a script?

> Some fast food places are replacing employees with an AI drive through.

What, as in it transcribes your order with Whisper and tries to upsell you through ChatGPT? One more reason I'm glad not to have a car-centric lifestyle, I guess. The kiosks inside the store might be vibe-coded now but at least I get a traditional UI that lets me specify things directly (even if the kitchen staff will ignore most requested customizations).

> Is this actually worse than being on hold forever to talk to someone following a script?

Yes. It means that common or sudden issues with the provider are not understood internally and huge amounts of customer time becomes wasted on a system with an out of date understanding of the service.

> as in it transcribes your order with Whisper and tries to upsell you through ChatGPT

Essentially. https://www.youtube.com/watch?v=zfz2EtWWPcQ

> One more reason I'm glad not to have a car-centric lifestyle

I'm always amazed at this logic. It makes me wonder if you either have an incredible amount of free time or you don't rely on any service provided to you with a vehicle, or are you just not considering them when pondering this way openly?

> at least I get a traditional UI

That has almost no accessibility for the disabled or has accessible functionality that's terribly tacked on as an afterthought.

> > One more reason I'm glad not to have a car-centric lifestyle

> I'm always amazed at this logic.

Not OP and maybe it's just my European showing, but I own a brand-new car yet frequently go 4-5 days without actually driving it. Because going to work and dropping off kids at daycare using a bicycle is literally faster than doing it in a car.

Whenever the issue of cars come up the gaping divide between Europeans and Americans rears its ugly head. It's like someone living at the equator telling the Alaska resident that though they own a winter hat, they rarely feel the need to use it. In many areas of Europe it is not only possible, but convenient to get around without a car. In the overwhelming majority of the United States it's impossible or inconvenient with only a couple of cities (like NYC) where it's even feasible. In fact, a growing number of eateries and coffee places in the USA are solely accessible via the drive through. They don't accept walk in customers at all and have no dining area.

This was a deliberate political decision by the USA. It wasn't an accident and it isn't because of geography.

But isnt the problem that america was built by americans to be car-centric? Most americans dont live in the middle of nowhere in a forest cabin or on a ranch. No one is saying individuals in america are bad for using a car when the system is broken. The critic is about the system.

> I'm always amazed at this logic. It makes me wonder if you either have an incredible amount of free time or you don't rely on any service provided to you with a vehicle, or are you just not considering them when pondering this way openly?

Not really sure what you're getting at. Yes, I would get in an ambulance, or even a taxi, if I really needed one. That's not what "not car-centric" is about. No, I don't need things delivered to me and I don't need a car to access goods and services. I don't buy a lot in the first place; public transit works acceptably here; I'm capable of walking several km (and I'd spend the time on other forms of exercise otherwise); I mostly cook my own meals.

And there are parts of the world where public transit is actually good and it's often rational to take it even if trip time is your only consideration.

> That has almost no accessibility for the disabled or has accessible functionality that's terribly tacked on as an afterthought.

Yes, I didn't say it was good. But you can also still just talk to a cashier at the front counter here.

I was being a bit of a harsh hipster there; however, I always think about the sheer number of vehicles required just to keep your electricity or high speed internet running and how often I see that type of vehicle in a drive through. I think it's sometimes a little easy to forget why our lives are as convenient as they are.

> the sheer number of vehicles required just to keep your electricity or high speed internet running

Yes, well, in large part that's due to choices other people could also make differently.

In places without widespread drivethroughs they still have electricity and internet.

Non-car-centric doesn't mean no cars. It means a society not centered around the crazy amounts of cars.

A lot of those types of workers who have vans full of tools use them as their main vehicles. That van being in the drive through doesn't mean the drive through is supporting the societal function that's advertised on the van. It just means a worker who does that for a living is currently buying food there.

> replaced by lengthy, egregiously pedantic AI systems.

Don't forget hallucinating too.

Amazon's hallucinating a fake 1-800 number for me to call is both peak Amazon and peak AI bubble.

Reply on news.ycombinator.com