Most consumers hate the idea of AI-generated customer service

Skye Jacobs

Posts: 44   +2
Staff
A hot potato: Customer service is one of those corporate functions that consumers love to hate: long hold times, incorrect answers, and less-than-pleasant interactions are all-too-familiar issues. AI could solve all of these problems, at least in theory, and save companies money to boot. But customers hate the idea, according to a recent survey by Gartner. The main reason is that they think AI will make it much harder to reach a human agent.

A majority (64%) of 5,728 people surveyed by Gartner in December 2023 said they would prefer companies not to use AI in their customer service. Additionally, 53% of customers hate the idea so much that they would consider switching to a competitor if they found out a company was going to use AI for customer service.

The survey is timely because customer service leaders are eager to use AI, with 60% under pressure to adopt the technology, according to Gartner senior principal Keith McIntosh.

The respondents said their top concerns were that it would become more difficult to reach a person, AI would displace jobs, and the technology would provide the wrong answers. Other fears are that their data would be less secure and AI biases would not treat customers equally.

"Many customers fear that GenAI will simply become another obstacle between them and an agent," McIntosh said.

One solution for companies itching to try out the technology in their customer service operations is to promise customers that they will be connected to an agent in the event that the AI cannot provide an answer, McIntosh said. "It must then seamlessly transform into an agent chat that picks up where the chatbot left off."

Consumer disapproval of AI use in customer service is unlikely to keep firms from deploying the technology as the cost savings are just too great. AI in general can automate routine tasks and inquiries, and this applies to customer service as well. To be fair to companies, AI can also be used to augment human customer service reps by providing them with answers faster than they could look up themselves.

But there are some human touches that AI simply can't fake for customers. AI is no doubt efficient, but it struggles with responding to complex emotional situations that require human empathy. And like human reps, sometimes it just flat out gets things wrong.

One recent example is McDonald's decision to discontinue the use of AI in some of its drive-thrus due to the errors it was making – errors that found their way onto social media.

Permalink to story:

 
I was trying to watch samsung reveal right now, but every sentence had an AI in it making it difficult to listen to, and paid crowd reaction goes woooooo every time any nonsense is said. I woke up - woooooo! , went to a car - woooo!, and so on.
Happy to see a new devices coming, but I cant stand that cringe stuff.
 
:rolleyes: IMO, AI for customer service is a totally horrible idea.

Whenever I have used a "bot" for customer service, anywhere with any company, all the "bot" has done was tell me to do what I already did before I used the bot. I always end up telling the dumb bot that I want to speak with a real representative. How dumb and cheap can you get???

IMO, companies using bots and companies that plan on using bots have only made, and will make, their already crappy customer service even crappier.

More power to them. If those companies want their crappy reputation for customer service to be even crappier, let them use AI bots. They will deserve the reputation they get.

In many cases, especially customer service, AI is totally useless, energy-wasting, crap, IMO.
 
I was trying to watch samsung reveal right now, but every sentence had an AI in it making it difficult to listen to, and paid crowd reaction goes woooooo every time any nonsense is said. I woke up - woooooo! , went to a car - woooo!, and so on.
Happy to see a new devices coming, but I cant stand that cringe stuff.
Sorry, my friend, wrong article.
 
The idea of a customer support bot is not bad, but the implementations usually suck. Here vodafone makes it almost impossible to get in touch with a human and their bot is useless most of the time. I had to switch to another provider, good riddance.
 
Using AI for the first tier of customer service is fine… as long as if the bot doesn’t solve my problem, I have the option of proceeding to a human “supervisor”.

I can’t remember how many times I’ve called my ISP and sat on hold for 30+ minutes just to get a minimum wage person reading from a manual who failed to solve my issue. Only after an hour of wasted time would I finally get transferred to someone who can actually do something - which was generally “a tech will be at your place tomorrow”.

If a bot can reduce that time , I’m all for it.
 
Large language models to replace the super dumb bots right now would be great. Even small Phi 3 models are light years ahead of the dumb bots. Once you train them on internal documentation from Confluence and the like, they should seriously be able to solve ~80% of issues if our internal help desk is any measure to go by.
This doesn't mean fire all the humans of course. It means that the humans will only be dealing with the actual unique issues, which is way less boring and a better fit.
 
:rolleyes: IMO, AI for customer service is a totally horrible idea.

Whenever I have used a "bot" for customer service, anywhere with any company, all the "bot" has done was tell me to do what I already did before I used the bot. I always end up telling the dumb bot that I want to speak with a real representative. How dumb and cheap can you get???

IMO, companies using bots and companies that plan on using bots have only made, and will make, their already crappy customer service even crappier.

More power to them. If those companies want their crappy reputation for customer service to be even crappier, let them use AI bots. They will deserve the reputation they get.

In many cases, especially customer service, AI is totally useless, energy-wasting, crap, IMO.
I disagree. I have been working with many companies over the past 18 months deploying AI solutions. Many are customer or internal employee focused chatbots. If you train the AI properly, it can do good things and improve your customer service experiences.
 
Using AI for the first tier of customer service is fine… as long as if the bot doesn’t solve my problem, I have the option of proceeding to a human “supervisor”.

I can’t remember how many times I’ve called my ISP and sat on hold for 30+ minutes just to get a minimum wage person reading from a manual who failed to solve my issue. Only after an hour of wasted time would I finally get transferred to someone who can actually do something - which was generally “a tech will be at your place tomorrow”.

If a bot can reduce that time , I’m all for it.
Exactly this. With a bot, I can jump past the 34 steps a human will insist they make you go through, even though you have told them you have done all 34, exactly the way they want it done. Sometimes I just need some info and talking to a human in, some cases, just results in them trying to sell me something I don't want. I think the problem is they don't train the AI with enough data. Fortunately, if they are using good AI, they can continuously train it with new documents and data.
 
I disagree. I have been working with many companies over the past 18 months deploying AI solutions. Many are customer or internal employee focused chatbots. If you train the AI properly, it can do good things and improve your customer service experiences.
I guess I've had the unfortunate, but seemingly not so uncommon, experience of dealing with improperly trained AI, or being a customer that actually has a clue about what to do before contacting CS.

It would seem that companies that employ AI Bots for CS think that everyone that contacts CS is clueless and would just waste the time of real CS agents.

Maybe a solution for that kind of customer, like myself, that has tried all the recommended "resolve it yourself" suggestions is to have a menu entry, when contacting CS, that is something along the lines of "I've already tried all your recommended solutions, and nothing worked." That would, IMO, go some length to solving the problem of wasted time whether its the customer's wasted time, the bot's wasted time, or the live CS agent's wasted time.

But instead, we get companies buying into the hype of AI being the fast, cheap, and easy solution to the world's problems, and really, IMO, only exasperating the problem.
 
Using AI for the first tier of customer service is fine… as long as if the bot doesn’t solve my problem, I have the option of proceeding to a human “supervisor”.

I can’t remember how many times I’ve called my ISP and sat on hold for 30+ minutes just to get a minimum wage person reading from a manual who failed to solve my issue. Only after an hour of wasted time would I finally get transferred to someone who can actually do something - which was generally “a tech will be at your place tomorrow”.

If a bot can reduce that time , I’m all for it.
If you get the sense that you are talking to a first-level CS agent and that agent is getting you nowhere, you always have the option of saying "I want to speak to a supervisor" at any point during the conversation. No one has to sit and waste their time wading through N-levels of "I've already done that" when talking to a CS agent.

That's an approach I take since I realize that first level agents are likely just reading from a script, and especially if they are telling me to do what I've already done.

I find that that approach puts the control of the situation back into my hands and takes control away from the useless CS agent.

If the company is decent about it, the company will do as much as they can to impress upon you that they are giving you great customer service.

Trust me, if you let the company control the narrative, they will.
 
If you get the sense that you are talking to a first-level CS agent and that agent is getting you nowhere, you always have the option of saying "I want to speak to a supervisor" at any point during the conversation. No one has to sit and waste their time wading through N-levels of "I've already done that" when talking to a CS agent.

That's an approach I take since I realize that first level agents are likely just reading from a script, and especially if they are telling me to do what I've already done.

I find that that approach puts the control of the situation back into my hands and takes control away from the useless CS agent.

If the company is decent about it, the company will do as much as they can to impress upon you that they are giving you great customer service.

Trust me, if you let the company control the narrative, they will.
Alas, try dealing with Bell Canada… they won’t let you talk to a “supervisor” until they go through their script first… kills me…
 
I had a totally AI generated chat session with Hoka shoes, yesterday and it was utterly useless. I gave up in disgust and contacted them by email instead.
 
Back