The grim prediction comes from K Krithivasan, head of Indian IT giant Tata Consultancy Services (TCS). The second-largest company in India by market cap, it has more...
Idk— Tesla directed autopilot to disengage the moment before a crash, to ensure that the driver was held responsible. I sadly doubt any corpo would pay for its mistake
Call centres exist because people can’t get the help they need by searching. Take away call centres, and you’re just making it more difficult for customers.
Uhh they’ve all been outsourced to India for ages now, and they’re effectively useless. You’ll get someone who’s worse than an AI at understanding what you’re asking and cannot deviate from a scrip because they have no training.
I haven’t used one in over a decade, if I have an issue it’s going as an email or a comment on social media.
Can't even describe how shitty meta support is. I ordered a quest 3 and some additionals, they mailed me everything except the quest and something else I didn't order. Obvious mix-up, yeah? Well it took 4 different support chats and 6 different "specialists" over a month to actually process a refund, and they were still somehow stuck on the idea that the courier missed a box. An additional box under the same tracking number as another box that was labeled 1 of 1.
We handle support in our company as part of our day to day. By and large the bulk of support is people simply leaning on us, rather than relying on common sense, or using the docs. Only a small percent is what would be considered essential.
However, each industry is different. This is just ours.
Generative AI could easily help the bulk of our support load.
We're experimenting with retrieval augmented generation for early inquiries right now. We get hundreds of inquiries that could be answered by looking at the website/docs and Q&A models with extractive or abstract approaches, or newer generative approaches are good at handling them.
Looked at four models last week, 2 vendors and 2 open source solutions, it's very promising. Very high accuracy with extractive approaches to simple queries, an email answering bot that links to our live website, along with an offer to talk to a real person could help us out a lot.
You'd be surprised how often we can automate a customers enquiry with ML (not even generative AI). Humans are still there as a fallback, but it's a way better experience to give instant help to the person if possible and then put them into the queue if they have a more complicated problem. Searching is not really in the same context as automating customer queries, although I guess it could depend on the domain to an extent.
If a request is simple and common enough that the request can be automated, then it is most likely something that I'm already pissed off about having to call in about since it should have been a feature on the company's website.
Oh, I've been called by my cell operator today. Realized it's a bot when it offered to upgrade my plan because I've used more than half of it. It's 28th of this month FFS! I asked whether that's what they mean and why would I need that, it just repeated in the same voice.
Remember back in the early days of the internet, when FAQs had frequently asked questions, and were updated in response to calls? Pepperidge farms remembers.
Well, the world is your oyster... Can't wait for yout chief automated officer.
edit: There was a similar hype back in the blockchain era, where people were trying to build decentralized organizations by making all the shareholders directly vote on every decision. Let's just say this model wasn't especially successful except for very rare circumstances.
I work for the largest contact center tech firm, and I'm sure it won't be this year. There are major issues to be solved. The companies that acted first had cars sold for $1, entirely new offers made up that they had to honor, and their bots made poems full of swear words about how shitty their companies were. I'm not sure how long it will take for gen ai to take over but some major issues need to be solved first and I don't see much progress being made on those.
Will the AI not hang up on me when I ask it a question that's going to take a long time to resolve and fuck up it's service metrics?
I've gone through like 3 service reps in a single problem because the call mysteriously drops after I outline the issue. "Could you hold please?" --- click
I promise you 1) this real-world event will fail to be scrubbed from the training data & 2) it will be regurgitated as a valid event that saves the company money.
There are already stories about companies being sued because their AI gave advice that caused the customer to act in a manner detrimental to themselves. (Something about 'plane flight refunds being available if I remember correctly).
Then when they contacted the company to complain (perhaps get the promised refund), they were told that there was no such policy at their company. The customer had screenshots. The AI had wholesale hallucinated a policy.
We all know how this is going to go. AI left, right and centre until it costs companies more in AI hallucination lawsuits than it does to employ people to do it.
And all the while they'll be bribing lobbying government representatives to make AI hallucination lawsuits not a thing. Or less of a thing.
On the other hand, are you implying that human call center workers are accurate with what they tell customers and that when they make mistakes the companies will own up to them and honor them?
I mean that's generally how it is now. If a rep lied to me then you better believe I'm talking to the manager and going to extract some concession. The difference is you can hold a rep accountable, dunno how you do that with AI
I will not be shedding a tear if people no longer have to work in such a soul crushing menial job. Fuck around and findout what happens millions of people lose their jobs all at once.
but anyway this is just some more ai hype stock manipulation shit.
First of all, it will make cold calling way, way worse. Time to ramp up restrictions, fines and other penalties for that kind of stuff.
When it comes to tech support call centers, some may actually improve. Not because the technology is so superior, but just because the current support simply sucks, and any change would be an improvement. And then they must actually work, i.e. solve the customers problems. On top of that, there is that case where an AI call center made expensive promises (IIRC if promised a car for $1 or something like that), and the judge made the company uphold this deal.
I've explained that it means "reduce by 10%" once, got a long rant that in the modern language it means what people mean when using it. Linguistically correct, but heresy.
But also first language information systems are a real problem in India especially with government services. These tools are gong to make it much easier to provide vital services and connect people to the appropriate services.
It's been a couple decades since I worked in a call center (tech support).
Are they still dominated by shitty ticketing systems that employees are expected to fill out while being on the call? I don't know if that was just an oddity of the call center I worked for or not. If I didn't fill out a ticket correctly we wouldn't get paid for the tech support, so management would get real upset if you didn't fill out a ticket correctly. There were like 400 fields to fill out in a ticket and you had to fill out about 15 of them just right; fill out one too many, or one too few, or the wrong one and management is upset.
Honestly, language models would do better filling out those tickets than they would handling the call. If an AI can't fill out the ticket, how can it solve an actual problem? It would sure make life for the call center employees better if all they had to do was talk instead of managing a bunch of tickets and paperwork using shitty internal apps. But who am I kidding. They'll probably find a way to make life worse for the customers and the call center employees and they'll make a profit, because that's how free markets work, right? Whoever makes life worse for everyone prospers.
I am with you. We should use the ai as a tool to automate or remove things that is frustrating or in the way of the actual goal to help the customers. Plus I don't think any model is good enough (yet) to act as tech support (they can use open ai if it was enough). I think ai is great as a tool tho. For example you can use it to go through a lot of documents of products, policies, other tickets and so on so the tech support person can find the relevant information faster. We can also use ai to create summerise of the call or take notes and so on. A lot of great potential to make everyone happier but I don't believe in replacing actual ppl.
really hope so - my work's ServiceDesk is based in India and they're barely functional - nothing gets fixed until the problem gets passed to an American IT person.
The fact that generative AI is being used as a means of large corporations consolidating even more wealth rather than attempting to free the working class from shitty, menial jobs shows that we're way the fuck off with how we conceptualize of "work".
This should be a good thing, but for lots of people this will suck.
I might be the only one that's kind of optimistic this will improve some of the cheapest call centers.
Some of them ... the people have such thick accents, don't get any local references, the connection is bad, don't know the first thing about the subject matter, etc.
I called my health insurance company one time because CVS said my vaccine wasn't covered there; the lady on the other end of the phone I could barely understand and I had to explain to her that CVS is a pharmacy. She still didn't give me any helpful information. Eventually via poking around the website or something like that, I found out my insurance company doesn't cover pharmacist administered vaccines
... which is just insane to me.
Doesn't sound insane to me, sounds like a perfect example of the state of the US healthcare system. I stopped paying for healthcare 2 years ago, best decision I ever made.
It's a revolving door anyway. I think the average length of time somebody works for a company in the industry is 7 months.
Besides the jobs that AI will take over will be the higher paid ones because inevitably that will result more value for money. Low wage employees are less of a burden for a company and so there is less incentive to replace them.
In one way, I'm happy this is happening, in another way, I'm not - I've given well over 2 decades of my life to the call center way of living. Let me give you a sneak peak into what really happens in the daily life of a call center worker.
You live by the time on your telephone, it's your punch in and punch out system in most centers. Don't clock in more than 8 or 15 or whatever insane metrics they set past your clock in time else you will be considered tardy. This includes all breaks and clocking out.
If you are a first contact person and taking phone orders, your 'talk time' is measured. Anything more than the standardized 5 or 6 minutes is considered excessive and they tell you to move the calls along faster.
If you are customer service, your talk time is loosened but you are also the first and last contact the customer should have for the issue.
Your phone calls are monitored and/or recorded (For Real!). If you are like me and hate to your your voice, woe be it to you when they play back your last call or two so you can hear yourself talking to the customer. If not recorded, then it is up to the monitoring person to be nice. You are then told what you need to do to speed up your talk time, or increase sales etc..
Telemarketing
Oh dear God, this is a life sucker and has the highest turnover on jobs. You quickly learn more about human nature in an odd sense. The sheer pressure on booking that next sale is insanely high and if you don't meet the sales minimums for the day or even hour, you are sent home without pay. I worked for a company which sold HR Manual trials, I was never more relieved and happy to be fired when I was for not making the per-requisite sales quotas for the half day.
TIPS
I don't think I've encountered a single call center rep in my years of service where a CSR decided that today, they would be a jerk. All we ever want to do is get through the day and earn our wages and go home.
One thing I will say with confidence, is everyone you work with has something in common, you aren't there necessarily because you enjoy it, you are there because it puts food on the table and beats living off of unemployment benefits. It's a thankless job.
If you receive great service from a call center rep (CSR) and are happy, politely ask to speak with their supervisor and when you do, be sure to leave them a good review. It doesn't always help to do this after a bad call, but sometimes rebounding to a new agent by calling the company back and asking for a supervisor will make a big difference if you take issue with them about the poor quality of service you received.
Remember, if you can't resolve an issue with a CSR, It's not always that they don't want to resolve the issue for you, their hands are probably tied and in fear of losing their job or being reprimanded, they simply won't budge.
Kindness goes a long way with us as well, if you are respectful and kind, we reflect the same back to you and often have tools at our disposal to grant you an extra discount and/or savings. We genuinely want to see you happy!
ON THE OTHER HAND
If putting AI in front of the call centers will help screen out the most common issues, then by all means do it. Also, if the stupid bean counters out there which insist of outsourcing to third world countries as it's cheaper, can find it to be more cost effective to use AI, and keep the jobs local to their country of operation, then I'm in favor of it.
It's already in the works. We leverage genrative AI via a chat bot in our internal ticketing system for ticket deflection and we're currently rolling out a similar feature for external customers as well. This goes well beyond simply linking high level FAQs. The bot asks a series of questions based on the issue and goes through the same line of questions our service desk reps ask to help diagnose and resolve the issue. And if they go through the entire T1 process within a minute or so and don't have an answer, it creates a ticket for the appropriate team based off of whatever platform, app, or website the issue is in regards to.
It's crazy scalable and has allowed multiple teams to shift focus towards more project oriented work. Without it we probably would have hired more service desk reps as the company grew
We still get some internal customers that just mindlessly click through it or slack us directly but that's the joy of IT.
I know this isn't a common takeaway, but I'm all for it.
The current state of call centers is EXCRUCIATINGLY painful for consumers and only exists in the form it does to commodify US for the company. We tolerate the shit experience of call centers, so companies don't need to pay more money to give us a better experience. That's why they exist, the only reason. If firing them all and turning them into AI makes the experience even SLIGHTLY less painful than calling your local public assistance help line, I'm all for it. If I can bypass 4 separate phone tree selection menus with 3 minutes of wait time with the crackling loud wait music between being passed around departments, I'm all for it. These are shit dead end jobs with no upside whatsoever.
I work for a company that is “all in” on AI. And offshoring. But AI is unlikely to provide second or third level support for complex and poorly documented software that operates at the intersection of legislation and rule making.
Add to that, customers who are licensed in their field but cannot comprehend that software implementation of paper forms requires the same inputs generally, much less explain their objective…
Also, the implementations I’ve been presented with as a consumer have been hot garbage.
The front line folks who exist primarily so customers can yell at someone might be in trouble. But companies who put their people in that position are shit anyway.