Hacker News new | past | comments | ask | show | jobs | submit login
SoftBank's new AI makes angry customers sound calm on phone (asahi.com)
73 points by achristmascarl 5 months ago | hide | past | favorite | 92 comments



This is only needed until the AI is good enough to totally replace the human customer service agents. And that probably isn't too far off. The AI won't care if the human on the phone is angry, since it doesn't have emotions.

There could be a silver lining. A big part of the reason that it's so hard to talk to a person when you call is that people are very expensive. If the AI is cheap enough, everyone will be able to speak to a "representative" without waiting on hold or navigating a menu. It can just answer immediately and say "how can I help you" and have a conversation. And if the AI is helpful enough, fewer callers will be angry, because their problem is being solved and they aren't spending hours on hold or being asked inane questions from a script.

Silver lining for everyone except those employed in call centers, of course.


> The AI won't care if the human on the phone is angry, since it doesn't have emotions.

I'd say that AI can care about humans emotions, but the wonderful thing is AI doesn't have to have programmed emotional responses to emotions of others. Theoretically speaking, AI could respond to anger by applying some soothing techniques, like acknowledging emotions and active listening.


I would prefer it would solve my problem instead.


Genuinely: Why do some people assume that AI customer service will magically have more power than human customer service reps?

In my experience, human customer service typically sucks because the person is highly constrained by the organization in what they can and can’t do and say.

What evidence is there that companies will give AI more discretion+power than current human employees?


My view is that you need human customer service mostly for cases you can't serve with a self-service UI. And if you can't build a UI for it, you certainly can't build an AI for it. So the AI is necessarily less useful than either the human or the self-service interface.

A lot of companies expect big savings from AI customer service, so where do I assume wrong here?


I agree with you that AI probably won't have better than human customer reps access to systems and solutions.

Then again, most customer service reps are so limited that I can easily see an AI having an equal amount of power and discretion on the things customer service reps already do. Really this would just be a push people to the next tier of support type thing in my mind.


> If the AI is cheap enough, everyone will be able to speak to a "representative" without waiting on hold or navigating a menu.

Sometimes people want to "speak to a representative" because they're too lazy to navigate a menu or search, and would prefer to use a natural language interface. AI will definitely help with that.

But for most people - especially younger people - our interface desires are not the issue. We prefer navigating a menu interface to having a conversation with anyone - AI or human!

We want to "speak to a representative" because the automated interfaces are deliberately designed not to let us do the thing we need to do. The customer service representative is not our servant, they're a guard who holds the keys to the door we need to enter.

Will companies give those keys to an LLM? I doubt it. It's certainly possible to trick CSRs, but it's easier (and more automatable) to trick LLMs. Instead, we'll get LLMs which claim to be CSRs but don't actually have any keys, bringing us back to the time-honored demand "I want to speak to a human!"


I had a package that I ordered on Amazon recently get lost in the mail. The order page on the website stated that I can ask for a refund now There was no obvious button to click to actually do so. When I tried to reach out to customer support, I found that Amazon had completely replaced their chat support with an AI.

Every single time I tried to explain to the AI that my package was late and I needed a refund, replied that if the package doesn't arrive by [date that was the previous day] Then I can come back and ask for a refund.

I kept trying to explain to the bot that the date it provided me has passed but it wouldn't listen and just kept throwing me through a loop.

In the end I had a request that Amazon call me and the representative solved the problem in less than five minutes.


> We want to "speak to a representative" because the automated interfaces are deliberately designed not to let us do the thing we need to do. The customer service representative is not our servant, they're a guard who holds the keys to the door we need to enter.

Nailed it. A lot of times you're doing something special or off the well beaten path, or they just don't want you to do it (like closing your account). Voice recognition usually just seems to listen to single words to try to route you to to someone. Either way, usually by that time the company doesn't care, they've got your money.


I wonder if the beaten path is a small minority of customers with repetitive issues, and therefore the majority of customers are off the beaten path when they call. A distribution of call types that is a small spike and a very long tail.


Yeah I usually want to speak to a human for the exact thing AI usually sucks at: a rare occurrence that I need someone to manually intervene in.


The AI chat bots dealing with written language are crap enough on customer support scenarios. Throw in voice and regional accents and it's going to be even worse. No thanks.


They should still expose the emotion level to the call center operator, just in a different way. Have something on the UI that indicates the level of emotion. It's still an important metric to take into account when resolving the customer enquiry.


The article addresses this:

…if an operator cannot tell if a customer is angry, the operator may not be able to react properly, which could just upset the customer further.

Therefore, the developers made sure that a slight element of anger remains audible.


That doesn't address it. That tones it all the way down to a "slight element of anger"


This makes the assumption that the emotional level of the caller matches the severity of the issue but I don't think that is often the case.


Perhaps not in reality, but it's the customer perception that matters.


The whole point of this tool is to help relax the employee. The customer perception doesn’t matter here.


1. Customer is mad about problem 2. Support person does not know this and speaks as if the other side is happy and casual 3. Customer gets even more mad because they feel their problem isn't being taken seriously, leading to some sort of escalation

Customer perception does matter.


And how does the script-reading, process-following, zero-power custech employee change what they can do based on customer rage?


They aren't robots, they can put on a more genuine tone explaining to the customer that they can't do much, or escalate to someone more powerful without the customer having to argue with them to do it.

A family member had a banking problem last year (they got locked out of their account due to accidentally tripping account theft protection). Of course initially they weren't all that mad, it was their own fault. But customer support staff just had them running around in circles, gradually getting more and more infuriated until one support person finally empathized with their months of daily struggle and bothered to escalate them to someone who was able to force through a fix.

I had a similar experience with my university last year, the support employees were constantly tone deaf about the serious trouble they were causing me by asking me to drop all my other obligations immediately to deal with the consequences of their mistakes.

Every "sorry for the inconvenience" just rubbed salt into the wound because it just further emphasized that they didn't give a shit and that I had no recourse. Especially because the employees did have ways of making things better, by being proactive in providing information and making the accommodation adjustments that were within their power without my having to ask about them.


This sounds like a much broader and uglier problem in the system than empathy between custech and the customer - namely, empathy between the broader company and the customer. I am not convinced that the kinds of companies that will be using these services are the kind of company that even listens to their customer support agents, and would rather sell them out to the lowest bidder and the only reason they have any customer support agents at all is so they don't get in legal trouble because their automated system doesn't have all edge cases supported.


Yeah I suppose that's a fair point to make.


Thr


Perfect. We'll just build an AI to translate things into a "genuine" tone for the customer. All problems in society will be solved by adding a layer of AI indirection.

</s>


I would think calm and casual support is probably the best way to deescalate.


Surely all of us have experienced having to talk to someone who is angry, and even if we can't do anything about it, adjusting the way we speak to show that we understand their anger?

Eg letting them (to an extent) rant about the trouble being caused to them, phrasing things to not come off as confrontational or blaming, anticipating things that might trigger another outburst (eg being asked to try restarting the router when they've already claimed to have called in several times and thus may have been asked to do so several times).


Obviously the AI needs to adjust the support person's voice to match the customer's, so the customer gets back what they give.

Prior art from art: https://m.youtube.com/watch?v=naleynXS7yo


> Good news for call center operators who are worn down by hostile customers with unreasonable demands

What about angry customers with reasonable demands? Can the AI identify that?


Then their demands will still be reasonable, their frustrated tone will fade. The words are still good.


I can't help but notice LLMs are producing voices that are each day more human-like and now we expect the uniqueness of human voices and what they're trying to communicate to just fade away.

At which point does it stop making sense to have a human at the other side? Just let frustrated humans talk to the LLMs. They aren't hurt by your frustrated voice.


There are still jobs that absolutely, positively require a human, and I fear very much for situations where I need a human to talk to to resolve an issue, but one doesn't exist.

As a particularly morbid example, most systems out there are not prepared, at all, for an account holder to be dead and other people need to resolve their matters. Navigating the more robust conversation trees to get to a human in those extreme edge-cases is a whole thing.


This article lacks context.

For the past year or so, the term 'customer harassment' has become a trend in Japan. It's basically customers abusing the overly polite service staff over very minor issues. This AI is introduced to combat this, as there is now a severe staff shortage in Japan due to the reducing population and companies are having trouble enticing younger people into these kinds of jobs.

Here's a good article explaining it.

https://www3.nhk.or.jp/nhkworld/en/news/backstories/3105/

> Tokyo saying no to 'customer harassment'

> Types of 'customer harassment'

> - Targeting individuals as a form of stress relief.

> - A distorted sense of justice, in which people express their feelings for a long time as advice.

> - Attack type, in which people demand things such as having a person kneel down on the ground.

> - Illusion type, in which people think they are passing the responsibility on to others.

> - Strong obsession, in which mentally unstable people are obsessed with listening to their own opinions.


This use-case may be good or bad, but the logic underneath it is 100% correct IMO. Fundamentally, these new models allow you to encode and communicate high-level thoughts in the same way that the internet allowed you to encode and communicate well-defined information.

The natural evolution of this technology is to insert it into human communication channels and automatically transform raw thoughts into something better for the other end of the channel. "Better" is open to interpretation and going to be very interesting in this new world, but there are so many options.

Why not a version of HackerNews that doesn't just have commenting guidelines, but actually automatically enforces them? Or a chrome extension that takes HN comments and transforms them all to be kinder (or whatever you want) when you open a thread? Or a text input box that automatically rewrites (or proposes a rewrite of) your comments if they don't meet the standards you have for yourself?


> Or a chrome extension that takes HN comments and transforms them all to be kinder

I was about to start working on something like this. I would like to try browsing the internet for a day, where all comments that I read are rewritten after passing through a sentiment filter. If someone says something mean, I would pass the comment through an LLM with the prompt: "rewrite this comment as if you were a therapist, who was reframing the commenter's statement from the perspective that they are expressing personal pain, and are projecting it through their mean comment"

I find 19 times out of 20, that really mean comments come from a place of personal insecurity. So if someone says: "this chrome extension is a dumb idea, anti-free speech, blah blah blah" , I would read: "commentor wrote something mean. They might be upset about their own perceived insignificance in the world, and are projecting this pain through their comment <click here to reveal original text>"


Also on my project list - the AntiAssholeFilter. It's so interesting the ways you could handle this. Personally, I would want to just transform the comment into something that doesn't even mention that that the commenter wrote a mean comment - if it has something of value just make it not mean, otherwise hide it.

A couple things are really interesting about this idea. First - it's so easy for the end-user to customize a prompt that you don't need to get it right, you just need to give people the scaffolding and then they can color their internet bubble whatever color they want.

Second, I think that just making all comments just a couple percent more empathetic could be really impactful. It's the sort of systemic nudge that can ripple very far.


I appreciate someone actually putting an argument for this, just so we can tease out what's wrong here.

Customer support is one place where I don't want to just "send information". I want to be able to "exert leverage". I want my communication to be able to impel the other actor to take action.

The thing with hn comments is that the guidelines are flexible, even things that violate the guidelines are a kind of communication and play into the dynamics of the site. The "feelings" of hn have impact (for good and ill but still important).


Customer support is interesting here. I think you're very right (personally I think that the new AI in the article is a bad idea). I wonder if the ideal is transforming speech into multiple modalities. Maybe make the speech kinder to avoid activating people emotionally and then offer something visual that indicates emotions or "give a shit" level. But I dislike speech as a modality; the single channel nature is incredibly limiting IMO.

For HN comments, I think you're right. But I think there is still lots of potential there, from tooling to help Dang use his time more effectively to tooling that you can switch on when you are in a bad mood that lets you explore your curiosity but filters out/transforms the subset of comments that you don't have the emotional capacity to deal with well.

The cool thing is that this tech can be easily layered on top of the actual forum (does vertical integration give you anything? certainly crazy expensive) and so the user can be in control of what filters/auto-moderation they embrace. Plus text makes it easy to always drill deeper and see the original as needed.


Adding additional meta-data (like subtext to the raw text) is a good idea - however rewriting and automatically transforming behind the scenes is a fantastic way to create an even larger potential miscommunication gulf between two parties.

Even an individual extension or system that automatically transforms data into your desired content risks creating an artificially imposed echo chamber.


> risks creating an artificially imposed echo chamber.

I think that ship has sailed? Agree that the ramifications of auto-transforming communication is huge, but I think I'm more optimistic. The internet is a cesspool, I think that improving things is pretty likely now that empathetic software is no longer the stuff of dreams.


Smarter writing tools might be cool, but thinking of this as “enforcement” is kind of backwards given the current state of technology. AI is gullible and untrustworthy and it’s up to us to vet the results, because they could be bananas.

I think proposing a rewrite and letting people decide (and make final edits) could work well, though.


I think enforcement could have some positive uses. Think of reddit automod and what you could do with prompt engineering + LLM-driven automod. People's opinion of automod will vary, but I think it is a powerful tool when used right - there is so much garbage on the big parts of the internet, it's helpful to be able to deal with the worst stuff automatically and then incorporate a human escalation process.


>but actually automatically enforces them?

[strike]Democracy[/strike]Moderation is non-negotiable.


> Moderation is non-negotiable.

Typically, yes, it is.


[What you type in the text field]

This new moderation system sucks, who the hell thought of this?!

[What is displayed to the world]

I love this new moderation system, makes the site so much better to use!

----

You then think, fuck this, I'm deleting my account. Which you do losing access to it. But you see all your content is still there, and new posts are being made daily. Welcome to the new AI world where humans need not even apply.

Will it get this bad, no idea, but I bet some site will try it.


Makes a lot of sense.

On the other hand, I think there will be a tipping point where if there's enough hints that what you're saying over the phone is not exactly what the other person is hearing, you will become even more frustrated / angry and will simply stop trusting support calls in general. Humans are too perceptive.


This is exactly what not to do. I'd expect customer sat and churn to be a catastrophe. Video calls with high stress customers is a much better and probably far less expensive solution.


I wonder if they’d detect my calm, angry voice, learned from my customer service mother as a child. It’s not popular relative to American rage but it’s been an invaluable life skill.


What is calm and angry exactly though? Can you share a sample video?


It’s a calm demeanor and a visibly upset affect. Anger is not rage, etc. I do not have any videos of being upset (and certainly HN is no safe place to share one regardless).


That technology will pair nicely with this: https://www.vice.com/en/article/akek7g/this-startup-is-selli...

What a dystopian future that will be, having a whole telephone conversation where neither of you hears the other's real voice.


> having a whole telephone conversation where neither of you hears the other's real voice

This is actually not the worst-case outcome, in my view. The silver lining on the horizon is that hopefully there will also be a consumer-side model counterpart to these things:

I'd love to be able to just write an email. If the company is smart and offers a support email address, they can just get the text directly; if they insist on wasting my time by having only a number to call, their bot (or employee, but I suspect this will become increasingly rare soon) can argue with my bot.


Maybe it will all be good when there's an AI assistant that I can command to "Call the Foobar Company and cancel my subscription" and it can figure out the deeply hidden customer service phone number, navigate the phone tree menu, and then talk to their AI service agent to get my subscription canceled. It could be marketed as the subscription terminator. "It can't be bargained with. It can't be reasoned with. It doesn't feel pity, or remorse, or fear! And it absolutely will not stop, ever, until your subscription is canceled!


I feel like this comment will receive some visceral reactions, but I wonder whether similar technology or even LLMs could be used in emergency dispatch in the future. There could be benefits in terms of accuracy, speed, information verification, etc. Not to mention it would spare some 911 operators of PTSD.


Not sure about realtime voiceover and not in emergency services myself but having an LLM fed the call audio to generate supplmentary summary of what was discussed could certainly be beneficial in situations where the operator needed clarity or there was a mistaken address or important detail that needed double-checking after the call.


> According to the company, the biggest burdens on operators are hearing abusive language and being trapped in long conversations with customers who will not get off the line—such as when making persistent requests for apologies.

> Nakatani said, “AI is good at handling complaints and can do so for long hours, but what angry customers want is for a human to apologize to them.”

An aside: is this something people really call up call centers for, for an apology? If I'm calling it's because I need something, or something is wrong and I need it fixed. I don't need a random call center person to apologize on behalf of a corporation. I just need my thing resolved.


I wonder does it work in real time or does it add latency after each sentence?


Classic. Instead of fixing the problems so there are no angry customers just make it so the angry customers can’t be heard. What could possibly go wrong.


There will always be angry customers. People who call customer service and abuse the representative are not expressing anger solely due to the complaint. It’s an accumulation of other things in their lives.


When I call my insurance company and their sound recognition is so sensitive that if I make any noise at all it starts over and says "i'm sorry I don't recognize that" and then it takes me 10 minutes to get someone on the call, all while I'm paying them $25 every day of the year (it will rise another 10% next year), and the call center is in another country so it's hard to understand them and the call quality is shit, it's not other things in my life.

It's a broken system to divert money to executives and shareholders (most of whom have done zero work for the company), resulting in shittier service and anger for all of us.

BTW, I don't abuse the representatives with cursing or name calling, but I can get upset and I do let them know it's not their fault.


> There will always be angry customers.

There will always be companies that make their customers angry. You're really going against people seeking solutions from companies they pay?

I'm amazed anyone would downvote this - how am I disillusion here?


The customer is not always right.


Please don't straw man. Nothing in OP's comment said they're against fixing issues. But they - correctly - pointed out that you won't be able to completely eliminate abusive callers even if you're fixing issues. They never once presented it as dichotomy.


I would argue the strawman is the depiction of every upset customer being irate because of a personal problem instead of the issue they're contacting customer service about.


> the depiction of every upset customer being irate because of a personal problem

That's the straw man. I don't see anything in the parent comment that implies or expresses that. It says

1. "There will always be a subset of people who act irate or abusively for external reasons", not 2. "All people who are irate or abusive due to external reasons"

Your flipping the implication within the sentence, then arguing against that


Agreed


I mean, have you ever called Comcast?


Does it? It doesn't change the content, just the tone. If I were a call center operator I might want something like this


The content is only a part of it. We are a species specially-suited to pick up on others emotional states and factor that in our responses. A mismatch in response to certain emotions is going to cause further problems as, to an angry customer, it comes off as patronizing, dismissive, and disrespectful.

Typically, angry customers are conveying severity and urgency. If you do not respond with a similar sense of severity or urgency you now have two people who aren't on the same page. That's not a good foundation to solve problems on.


There's also good angry, where you can tell the person is angry, but not at the call center person.

I personally take care in this, saying things to reassure, such as indicating that I know it's the company's fault, not the rep's. This vecomes more of a "bartender" situation, where the anger is clearly at the situation, not the rep, and can create a more engaged, helpful response even.

I think one could sue for purposefully changing a person's appearance or voice, without their consent. And no, calling a help line can never ever have a tos enforced, as often the entity is required by law to deal with warranty or other issues. No tos applies.

Back to changing the voice, as others have mentioned, tenor, tone, etc all are part of speaking. They cannot be separated.

Tone is often, for example, brought up in testimony as it conveys something. It has as much meaning as the words.

I would be astonished if this isn't challenged soon.


As someone that's worked in call centers for a fair time in my life, I'm not sure about this.

I've worked for a cable company for a while that really turned to shit, and went into buying off representatives rather than fixing their technical problems (heavily investing in making city owned fiber illegal in many states). Customers being enraged with us was a very strong motivation for me to leave and find better work with a less abusive company. And yes, call centers commonly shut down/move/ or actually start giving better service when they can't keep staff because they are such a crap company that their users are in constant rage mode.


Filtering and distortion of communication. Suits well the age of appearances we live in.


I wonder what would happen if the customer is very happy and having an enjoyable conversation. Would the AI remove the emotions here too, resulting in a completely monotonic voice?


Has anyone found a demo of this thing? I'm super curious


The video on this page has one brief example: https://newsonjapan.com/article/142113.php


Sounds like something that can be applied to the call center operator voice as well.


I wonder how this AI interprets customers who express anger and frustration through sarcasm


<pushes AI button> "These aren't the droids you're looking for"


I like this use case. - Even if not entirely practical, it highlights a real issue and challenge we face as a society. Let's try to auto-tune down the tone of angry customers.

Our brains are wired to connect with others, and this includes picking up on and replicating emotions and behaviors. It's part of our mirror neuron system, which helps us understand and empathize with others. While this is beneficial in many social interactions, it can lead to escalation in conflict situations if not managed properly.


Sounds like a great use case for AI. Dealing with angry customers all day must be hell for one's mental health.


Are there any videos/audio examples of this in action?


I guess it’s always April 1 somewhere in the world


I would be extremely weirded out if I took a really angry tone with someone and they completely ignored it. It would probably be infuriating to some too if they continued sounding really chipper and positive and the customer is really upset. It might have the exact opposite effect and customers might just start saying toneless but very mean words to get their point across.


It's already kinda lika that though.


ubisoft s gonna buy this customer support tech before anyone else


Failing to acknowledge the very real emotions your company has caused a customer, who is paying you. Classic!


You can acknowledge how customers feel while shielding your customer facing staff from verbal abuse. If you've ever worked a customer facing job you'll understand.


You're not necessarily wrong, but I can't help but think that it may cause further frustration for me if I'm trying to convey a certain sentiment of frustration (for the sake of the argument, let's say that I'm not being an outright asshole at this point) that may be lost on the rep that I am talking to, resulting in a misunderstanding, which would further irritate me and escalate the situation.

There's genuinely something to be said about honesty and accurately understanding someone.


so now the actual content of what the customers are saying will adjust to be more harmful to compensate.


HN's new AI makes angry commentators' comments sound calm.

Wonder how people here would like that.


I think I facetiously suggested that once, that all comments on HN should be run through an AI filter and rewritten for maximum civility, technical correctness and factual accuracy, with the most common branches of conversation being pregenerated to save time. And then eventually the entire forum could be automated, with nothing but an AI crawling the web, posting and commenting.

It would be both terrible and hilarious. I still kind of want to see what it would look like.


Maybe just put your customer support agents into massage chairs?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: