145
Klarna says its AI assistant does the work of 700 people after it laid off 700 people
(www.fastcompany.com)
this post was submitted on 28 Feb 2024
145 points (97.4% liked)
Technology
59436 readers
3359 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Citation needed? We're going to take Klarna's word for it?
The LLM worked better than the Indian service agent I got to talk to, however after I got an agent speaking my language things resolved
Newer ever take Klarnas word for anything. They are the fine and Dandy company whose business model involved by routine fishing for customers bank authorization credentials.
Trust the robot. Robots never lie. Also I’m not a robot. Beep
I actually believe it.
Customer support for cheap companies is someone in a call center in India or some other country with obscenely low wages, following a bad flow chart with limited access and less autonomy. To match it you basically just need the LLM to parse the interaction into following the flowchart, and even when it struggles to do so, you're comparing it to a minimum wage, probably uneducated, worker in a non-English speaking country.
The bar is wildly low.
So "customer service" (the bastard child of huge companies and low wages) is actually just a really bad product to start with, and now Klarna has just replaced it with something that functions just as bad? Yeah I think that makes sense.
To quote flamingo_pinyata: "Zero equals zero"
Watch their stocks. 🤣
I would really appreciate it if anyone could share some video or article or something that explains the relation between stocks and layoffs.
I don't nave a citation, but in general, layoffs are usually used to cut costs. Spending less means more profits. More profits generally means the company looks better to the investors, and hence, better stock price.
I'm only taking about the relation between stocks and stupid decisions.
Wish I understood what drives the yearly layoffs tide
Companies will time layoffs to get a better profit in the next couple months to report better quarterly or yearly earnings reports. How those earnings reports turn out directly affects the stock market performance, which in turn makes the shareholders significantly more money.
This is most effective if somebody's trying to pump the stock value before jumping ship in the most egregious cases.
Often companies have layoff packages that pay workers X months of pay as a severance agreement, doesn't that mean they would be paying triple wages for some number of employees? Wouldn't that bring their costs up?
Good points!
The timing is quite important. Other things to consider are tax periods, bonuses, and nature of the markets. That can all be racked up as cost of doing business if the long-term benefits outweigh the long-term costs.
Especially if they are having a bad year or quarter, performing layoffs can show promise of a better next quarter since severance is basically a fixed cost to the number of employees you have.
There isn't necessarily one size fits all but the bottom line is dropping employees saves money as human resources are always one of the largest costs of operating.
Thanks for the explanation
"I decide what the tide will bring."
-Nami
So basically it's because of a fish girl.
What could they possibly cite?
"I'm glad we could solve your problem. Please rate you experiance with our customer support today"
Closes support chat window
Klarna: "Customers rate the AI assistant the same as the Human support workers" [all non-responses]
@Lmaydev@programming.dev exactly this in the comment above!
When Klarna says "customer ratings of the AI are the same as for human workers", what on earth do they mean?
It also contradicts the experiences people have been having with GPT powered chatbots recently. A model that takes a few prompts to start hallucinating is better than a living breathing human being? Really? I'd be curious to give it a try.
I worked for large companies before, and I even interviewed for Klarna (they have been hiring lots recently, maybe for different tasks/positions), and they always lie. Always.
At one company I worked at before, some KPI was calculated incorrectly and had been for years. When we informed the relevant person of this, they got very defensive and refused to change it. Only our team knew the calculation was total BS. It was become success/money was at stake for him. This person continued to send the fake KPI calculations to everyone every week.
I got laid off from the company where I work with absolutely ZERO motivation as to how me being laid off would increase "efficiency" (even though that was their claim
they are slimming down the company to increase "efficiency"). The company ignored every piece of evidence I provided to show that me and my team are completely overworked and that they should probably cut jobs elsewhere if they don't want all of our data ingestions and ML models to collapse.
I don't think companies have enough to motivate such a layoff if they can't solidify their numbers and make then transparent to the public. I'd certainly like to see the union fights Klarna will have with the Swedish worker unions now. I have never been to such a meeting though, but I've provided motivation for the union to use to oppose layoffs in such meetings - my understanding is that it's a bit of a shit show.
Yeah… we’re having this with one of our suppliers right now…
We’re showing them all the issues they aren’t solving and they’re saying “look all our KPI numbers look amazing”
So now we have to explicitly go into each ticket and mark it as dissatisfied, since they don’t take into account how long the ticket was open, how many meetings had to be called over the ticket, etc. just whether it was closed without clicking the extra “we’re dissatisfied” button
That sucks. A rigid KPI that is never open to change and only made to show some BLING in meetings, it will always lead to failure.
And when the KPIs are bad, they blame us. We had lots of failures in streaming videos on our platform and it was growing across a range of mobile OSs and devices, and they would just say, very sternly, "We need you to deliver better results!". Meanwhile, they had us do a huge migration and rebuild the entire UI across all devices all while maintaining the legacy systems that are sometimes riddled with bugs. Fucking idiots! They shouldn't expect X performance with Y² the number of tasks (I would have said 2*Y if it was "double" the work... it was more like being on steroids, it was Y² the work, everyone was burnt out, people barely took the summer off). We met those KPIs by a margin previously, and we're not about to meet them now unless they hire more staff or give us more time until their new UI launch. Spoiler: the launch did not go well.
Yeah, that’s always been my experience with KPIs, too.
It’s always turned into a game for managers - they don’t have the ability to actually improve the product, so the measurements change or they add an extra field in the ticket…
My first job was as a cashier and we were trained to press the calculate button on the register after each item we scanned. Did that make the checkout go faster? Nope. Did it make their measurement look better? Yeah, since that wasn’t “checkout time” in their report
At another job, patches weren’t merged fast enough. So, instead of making any changes that would improve the product, we were ordered to make one ticket for investigating the problem and a second for integrating the fix. It made fixing issues slower, but their KPIs went up…
You assume support agents don't hallucinate
🤣 yeah true
I feel like customer service has been on a decline for the past 10 years or so. I owe it to being underpaid and treated badly as an employee. .
Well clearly they should've done two decades long peer reviewed studies and even then we'd scoff and say they're obviously corrupted by *waves hands* you know, money, capitalism, you know, the system
700 people deserve some kind of study at least before their livelihood is cut off.
I'm sure they have some metrics of their own they use to follow the efficiency of their workers that they've used to compare
I'm sure your blind trust in Klarna is totally reasonable.
It's not blind trust to think companies probably try to make the most money out of the available resources. It's sorta what they do
Okay, just to be clear here, we're talking about the belief that Klarna has sufficiently motivated this decision to lay these people off and actually has good KPIs that measure the performance, specifically that of customer care agents... not that companies make money with resrouces.
Usually people with this kind of opinion like yours maybe haven't experienced work at a large company or maybe don't understand that office politics are alive and kicking. But I'd still like to hear why... Let me ask you, what makes you think Klarna is being honest about their measurements?
You don't think a big company like Klarna has metrics to follow the efficiency of their workers? Even small businesses where I live that and every single big corporation. What makes you think Klarna would be different?
You misunderstood. I said try to make most money out of the resources, so efficiency. That's what we're talking about...
Lol.
It's not very hard to believe what they're saying here and they'd be the ones to have those metrics. So simple 1+1.
I never said they have no KPIs. What I said is their KPIs are likely broken, just like almost every other large company in existence.
I think you're the one who misunderstood. So for reference again, I am talking about this: about the belief that Klarna has sufficiently motivated this decision to lay these people off and actually has good KPIs that measure the performance, specifically that of customer care agents…
I don't see why or how I would want to discuss that a company is an entity that makes money through products from resources... think that'd be a bit too basic, no?
?
But can you explain what makes you trust them? What gives a company like Klarna a high trust ratio in your eyes? They don't seem to have provided those metrics in any way, just spelling it out in words. So how can we trust them?
Maybe not for you, but the rest of us are skeptics and would like to know what exactly makes this easy for you to believe. Thanks for explaining in advance.
You started the discussion off from my joke comment saying that they deserve a study for them to be laid off. Consider where this started.
butwhymalemodels.avif
I just explained why their claim isn't hard to believe. You even quoted it??
Well no shit, not like I'm claiming to speak on someone else's behalf
Sorry, let's pretend I'm a total idiot, could you quote it for me?
Easy to believe them when the claim itself is easy to believe and they'd be the ones to have those metrics. So you take the easy to believe claim (1) and you take them being in the position to have those metrics (another 1, but let's call it 2 to differentiate) so simple 1+2.
Ummm, sorry but how is this an explanation?
"Why are they easy to believe?".... "Oh, it's just not very hard to believe, they have the metrics, they said so! 1+1!"
It's easy to believe because they are a big company and big companies have metrics?
But, what about the trust part? How does that make you trust them? What part of that is "enough" evidence for you to rationalize that they have a good motivation for laying off 700 people?
It's easy to believe because the claim (AI replacing the work of tons of workers) is easy to believe and out of anyone they'd be post positioned to know how well the AI compares to their workers. So not much of a leap to believe them here, that the AI can, could or has replaced the work of 700 people. Klarna seems to have something like 7000 workers so 10% being replaced by AI, for a company like Klarna? Yeah, I can see that happening. I'm honestly not sure why you are having such hard time with this. Maybe you don't understand those two as separate things, the claim (AI could replace a ton of workers) and the data the company would have (knowing exactly how many it could replace).
You've been the only one talking about trust.
Hmmm, okay, but let's take this one step further, what makes you believe that an AI can replace 700 customer care agents?
I have skepticism about all large companies, especially having worked for many, and especially having seen how far they will go with lies to lay people off. Klarna lays off hundreds of employees as a frequent occurrence.
In my opinion, both as an ML professional and as a chatbot professional who has built such systems or finetuned them for such tasks, it's extremely unlikely that chatbots would be able to replace 700 employees unless those employees were just doing simple zombie tasks.
The only sentence we get from Klarna in the article is this:
But nothing to elaborate. What was the satisfaction rating prior? No idea. So excuse my doubt when I don't want to believe a large company that lives to make money.
The idea of AI fully replacing a large workforce seems to be a bunch of BS driven forth by big companies like Klarna. Others in the field seem to be more careful about what AI can "help in" vs "replace". Have they A/B tested this? What is their KPI? Can we trust that KPI?
I invite you to read the rest of the comments under the thread. I think you're wrong.
Simple job + having ton of examples of it already + it's just another case of automation taking jobs that has been a thing for centuries + predictions from 4 years ago that automation would take away one third of jobs by the end of the decade and so on.
The number doesn't really matter if they're fine with just achieving the same result with less workers (less money). Customer care especially in English is already dogshit so AI achieving dogshit results is not hard to believe.
It's customer care we're talking about. Chatbots have already replaced a lot of it. First they outsourced to the who knows where and didn't care if the customer service sucked, now they just give it to AI and don't care if it sucks.
Okay, thanks. I think I understand your opinion better now.
I guess we disagree here. If the number doesn't matter, then they could just make it all up and move on. Seems like Klarna had no real number to share anyway, and depend on people believing that "it's easy for AI to take over any such job".
Yeah, but only to its detriment. Tricky problems go unsolved and chatbots are used by companies to create a wall between themselves and their clients. Revolut is one example that comes to mind. If customers cannot ever even reach a human being to tell them what's going on, then there's no problem!
I suspect Klarna is doing a similar thing. I also suspect that the worker unions in Sweden will not be satisfied with this reason for layoffs. Not sure if these layoffs are taking place globally or in Europe.
I meant that the number doesn't matter as long as it's the same number for both. They're talking about differences between the two and that's really the main point, whether the AI is any less shit than the people doing the job.
It's customer service. A lot of it has been taken over already and the rest outsourced to the lowest bidder in the third world. It's a very simple job without not much qualifications and it's very formulaic. Only requirement seems to have been some language skills and let's be honest, even that bar was very very low. So AI swooping in and making those jobs obsolete, I can see that.
A switch from a human insisting you reboot your router because that's what the sheet tells them to say to AI insisting you reboot your router because that's what the programming tells them to say, I can see customers thinking it's equally dogshit and since cost is the #1 thing for these companies, might as well go with the cheaper option. It's not like they cared about the quality of the customer service to begin with, otherwise they wouldn't have run it to the ground before this.
Mind you, if these were 700 highly professional workers fluent in English (or whichever language we're talking about) with deep knowledge of the problems their customers face and high social and problem solving skills, it's a harder claim to believe that the ratings would be the same. If it was your average customer service where someone who is paid nothing, who is given ridiculous performance goals, who barely speaks the language and just goes through the checklist without any care for your individual problem, with no actual knowledge about the situation, it might even change for the better.
But they probably are... or a large chunk of them. I disagree with that customer service is a useless job with no skill set.
Here's another article on the same topic btw:
https://gizmodo.com/klarna-ceo-ai-chatbot-customer-service-700-jobs-layoffs-1851293200
Seems like they decided this KPI two years after laying people off lol
To me, it looks like Klarna is just full of shit.
I really, really, really doubt they are considering what the situation with customer service of large companies is. But I suppose without either of us knowing more about the nature of these customer service workers, it's just a agree to disagree situation.
It's not useless job at all or rather the service customer service is supposed to deliver isn't. It's just that often large companies have cut costs so much on customer service that they're less able to provide a good service, to such a degree that it starts to feel useless and something you might as well replace by some AI.
Basically the point is that companies have made it so that their customer service is that robotic and shit and useless that you wouldn't even notice if it was replaced it with robotic and shit and useless AI.