Airlines Uses AI to Fake Empathy Rather Than Fix Problems: Passenger Sent Prompt(viewfromthewing.com)
viewfromthewing.com
Airlines Uses AI to Fake Empathy Rather Than Fix Problems: Passenger Sent Prompt
https://viewfromthewing.com/airlines-are-using-ai-to-manufacture-empathy-instead-of-solving-problems-one-passenger-was-sent-the-prompt-by-mistake/
16 comments
This looks more like "Human working at airline accidentally pasted their copilot prompt into whatsapp"
I hate how every support tool now has an additional “escape the llm” minigame at the start.
To be honest, I'm actually slightly sympathetic with the companies here.
My brother used to do tech support for XBox Live. He said that 80% of his calls were password reset requests, something that everyone can easily self-service, yet some people insisted on speaking with a human for it.
And it's not like the work flow was any different. He'd just get their username or e-mail and trigger the password reset e-mail, and they were happy and that was the end of the call.
I was at my father-in-law's place once and overheard him calling his cable company to pay his bill when he could have done it online in a fraction of the time.
Sure, there are people with weird edge cases that definitely need a human, and you should be able to get one when you need it, but I forgive them for making it slightly difficult when so many people insist on a human when they don't.
My brother used to do tech support for XBox Live. He said that 80% of his calls were password reset requests, something that everyone can easily self-service, yet some people insisted on speaking with a human for it.
And it's not like the work flow was any different. He'd just get their username or e-mail and trigger the password reset e-mail, and they were happy and that was the end of the call.
I was at my father-in-law's place once and overheard him calling his cable company to pay his bill when he could have done it online in a fraction of the time.
Sure, there are people with weird edge cases that definitely need a human, and you should be able to get one when you need it, but I forgive them for making it slightly difficult when so many people insist on a human when they don't.
Except that the same shit happens on ubereats/instacart/doordash driver apps where basically everything is self-serve except for a few critical interactions which basically always require a human. I can assure you no driver wants to talk to a human instead of pressing the button in the app. Still have to go through the same “no, it's not that, please connect to a human. No, it's not that either. No, you're not helping, please get a human agent” every time something unusual happens.
> Except that the same shit happens on ubereats/instacart/doordash driver apps where basically everything is self-serve [...]
That the password-reset was self-serve yet made up 80% of calls was Sohcahtoa82's point. Users that only call in unusual/exceptional circumstances can, by their nature, make a small minority of calls despite being the majority of users.
That the password-reset was self-serve yet made up 80% of calls was Sohcahtoa82's point. Users that only call in unusual/exceptional circumstances can, by their nature, make a small minority of calls despite being the majority of users.
I don't think you read my comment correctly, or misunderstood something.
The entire experience of fulfilling an instacart batch is a series of self-serve interactions; anybody who struggles with the basics won't be finishing the background checks to get started with the app, let alone completing enough batches to get out of new-user jail.
It is not an uncommon occurance for someone to order a single thing, and that thing to be out of stock, but such a scenario requires dancing with the llm every time before finally getting to a human where you can start the process of getting an impossible-to-complete order cancelled.
The entire experience of fulfilling an instacart batch is a series of self-serve interactions; anybody who struggles with the basics won't be finishing the background checks to get started with the app, let alone completing enough batches to get out of new-user jail.
It is not an uncommon occurance for someone to order a single thing, and that thing to be out of stock, but such a scenario requires dancing with the llm every time before finally getting to a human where you can start the process of getting an impossible-to-complete order cancelled.
Seriously. Ran into this with Tincan's support text line when I failed to receive the account initialization email to my provider...was immediately launched into LLM nonsense which I bailed out of until a few hours later when a real human finally texted me back. To add insult to injury, it turned out they couldn't figure out how to send emails to my small, obscure email provider, called 'fastmail' (/s) and insisted I use a different account. One failure after another with them.
In a way I found it useful because it showed me right away that they aren't really a serious business and I was able to cancel our account before the free trial was over.
(It also turns out that their base product is more or less a standard VoIP phone that is fairly easy to set up yourself for far cheaper...but I digress)
In a way I found it useful because it showed me right away that they aren't really a serious business and I was able to cancel our account before the free trial was over.
(It also turns out that their base product is more or less a standard VoIP phone that is fairly easy to set up yourself for far cheaper...but I digress)
Phone support robots had the same "escape the tree" game. You could usually bypass it and get to a human employee by spamming the zero 0 key.
This even led to a product: GetHuman
CVS Pharmacy has started rolling out an "AI assistant" phone tree with no apparent way to get to a human.
Their old school phone tree didn't either. You had to pretend to be irate.
Pretend? Oh, there's no pretending involved.
Maybe if you use a lot of profanities and threaten to cancel your subscription?
Yea. A lot of them grew wise to this and just hung up on you. Some had no way to get through the tree to a real person. Or, the real person couldn't do anything (e.g., Public Storage).
I think this is a bit too premature, but we really think that we will be able to talk to any sort of human when it's about support or contact? It doesn't scale, it doesn't make sense from a business perspective, look at how many new websites just don't even have a way to contact, this is strategical. Let's not forget that LLMs are best at language, they WILL be better than humans at speaking.
There is also a point where people will prefer talking to AI instead of human due to the pseudo-guaranteed accuracy, it's already happening within dev communities and many others. I personally prefer arguing with multiple models over my specs than another teammate, and this is the new normal already.
It's sad, but I wouldn't expect past a certain point to ever talk to a human when it's something virtual, there is also the perks of it which is customer not waiting, that annoying moment when you call HSBC and you need to wait 30min (day partially ruined) to get a human on the line which anyway will trigger all sort of automated/AI procedures, what's the point of the human at all?
Let's also not forget that humans interacting in call-centers don't have a lot of knowledge in general, you can see the requirements in the relevant countries (Philippines, India...) which are quite low, many people working in callcenters barely know how to use a computer or understand finance in any meaningful way, yet they work at JP Morgan, that's real life, I would prefer to be helped by a well-managed "AI" than an incompetent employee personally.
---
The general catastrophe in jobs and social issue is a different matter than accuracy, and even a different discussion. But for efficiency, it sounds absurd to think that AIs will not beat humans on a phone call/chat.
There is also a point where people will prefer talking to AI instead of human due to the pseudo-guaranteed accuracy, it's already happening within dev communities and many others. I personally prefer arguing with multiple models over my specs than another teammate, and this is the new normal already.
It's sad, but I wouldn't expect past a certain point to ever talk to a human when it's something virtual, there is also the perks of it which is customer not waiting, that annoying moment when you call HSBC and you need to wait 30min (day partially ruined) to get a human on the line which anyway will trigger all sort of automated/AI procedures, what's the point of the human at all?
Let's also not forget that humans interacting in call-centers don't have a lot of knowledge in general, you can see the requirements in the relevant countries (Philippines, India...) which are quite low, many people working in callcenters barely know how to use a computer or understand finance in any meaningful way, yet they work at JP Morgan, that's real life, I would prefer to be helped by a well-managed "AI" than an incompetent employee personally.
---
The general catastrophe in jobs and social issue is a different matter than accuracy, and even a different discussion. But for efficiency, it sounds absurd to think that AIs will not beat humans on a phone call/chat.