Normally I share tools that make your life easier, but today’s post contains anecdotes about how technology can make everything harder.
I have had it with AI chatbots in customer service, and I know I’m not alone.
Check out these stats from a SurveyMonkey report:

Although this was from 2023, the skepticism and negative feelings remain.
More insights from a 2025 survey:
- 75% of consumers say they’ve had a fast AI-driven response that still left them frustrated.
- Customers prioritize outcomes over speed, with 68% saying “Getting a complete resolution” is the most important in support interactions.
- Nearly 90% report reduced loyalty when human support is removed.
- Only 7% of customers say they rarely or never have to repeat themselves when switching channels.
- 34% say AI customer support “made things harder,” and a majority prefer “human-first” support pathways.
- 44% of consumers always try self-service first, and another 50% sometimes use it, showing strong appetite for AI-enabled resolution when it’s designed well.
If you’re here just for the lessons I offer, here are three takeaways from this post if you’re looking to outsource your customer service to AI:
- Don’t structure your AI agent feedback to perpetuate endless loops!
- Make sure customers can get to a human!
- Don’t let your AI agents pretend to be human!
Now on to my personal experiences.
I got my first credit card at age 16, and I’ve never had to initiate a credit card chargeback.
But recently I had to challenge transactions from two companies because I couldn’t get past their AI chatbots to fix billing problems. And another company almost lost tens of thousands of dollars from me because of their AI chatbot deception.
Let the ranting begin….
Quick Links
Don’t structure your AI agent feedback to perpetuate endless loops
The first one was my fault. I didn’t cancel an annual subscription fee in time for VideoGen, an AI-video generation tool. I caught it within 6 hours, and I wrote to the company to ask for some grace, especially since they didn’t send a reminder of the upcoming charge.
I went back and forth with a rep for a while, and she kept saying the exact same things:

It wasn’t until about the 8th email that I noticed the light gray signature line:

I got so frustrated that I couldn’t get any other answers or any way to connect with anyone. As a last resort, I asked my credit card company to reverse the charges. An important note: I take responsibility for not canceling in time. I would have probably just taken the loss and moved on, , but the fact that I couldn’t get any real person to talk with me was infuriating!
Make sure customers can get to a human
The second case was much, much more infuriating.
I paid for the pro version of Perplexity a couple of years ago, and then I saw that I got a free version through my Xfinity account for a year. I decided that I didn’t want to pay on my own for the third year (just don’t use it enough), so when they sent me this email asking for me to take action to renew, I did nothing.

It said very clearly that I had to take action, right? And I didn’t do anything, right? So you can imagine how angry I was when I saw not one but TWO charges for the same service.

Now I’m really mad (and confused). So I go to Perplexity, and a popup offers me a renewal at 50%.
What?
I composed a clear message to their customer service email that spelled out three things:
- I did not take action to renew.
- I got a random offer for 50% off renewal.
- I was charged twice for renewal.
Twenty minutes later I get an email from “Sam.”

To be fair to “Sam,” he/she did a good job of summing up my issue and sharing next steps. But look at the last paragraph.

So “Sam,” you’re telling me that if I reply and try to clarify what you wrote or even just say, “Thank you. I await your reply,” that you’re going to stick me in the back of the line again?
I seethed in silence for 5 days waiting for “Sam” and team to respond, but then I couldn’t stand it any more. I put in another chargeback request and wrote my AI buddy back:

I guess they really did move me back in the queue because it was almost a month after that before real-sounding person named Lilia responded.

Don’t let your AI agents pretend to be human
The O.G. Nerdmobile, my bright orange Prius C, wasn’t as reliable as we needed it to be for trips to/from the hospital for my hubby’s cancer treatment. So I found the car I wanted and contacted the dealership.
“Jerry” was so friendly and helpful. I looked forward to meeting him.





I walked into the dealership and asked for “Jerry” at the front desk.
The receptionist laughed at me and said, “Oh, Jerry is our AI guy.”
Well hell.
“He” had me 100% fooled. I thought for sure I was writing with a real person.
I felt foolish for being duped, and I resented the receptionist’s dismissive laugh. I’m about to spend tens of thousands of dollars with this company, and they couldn’t be bothered with giving me a real person to talk to? If I had more patience for buying the car I wanted, I would have walked out. It just turned me off.
If you’re using an AI rep, tell me. Don’t pretend it’s real.
Bonus Insight: Here’s how to use AI in customer service
I was kind of embarrassed at having to do chargebacks, especially since one agent at my credit card company said, “I see you have been requesting refunds a lot lately.”
Not a lot… just twice in my lifetime.
I asked the rep if he was a bot, and he laughed. “No. I’m real, but I get that a lot.”
I scrambled to justify my chargeback requests, and he surprised me by saying he had seen this frequently with companies who use AI chatbots for customer service. So I asked him if Chase credit card services used AI chatbots, and his answer showed the right way to use AI in customer service.
Because “all calls are recorded,” the rep said the company uses AI to analyze the call content and provide summaries and updates for the humans in the loop. That means that I don’t have to repeat the whole story to every new agent. They can see what I’ve been told and where things are now. This cuts down on both the length of the calls and the frustration of the callers, as well as the miscommunications.


Well said, Beth Z! Thank you for putting in writing what we have all been feeling lately.
I’m continually frustrated by AI chat bots for customer service, but I had an amazing experience with the AI customer service bot for my Oura ring. My ring is about two years old, so out of warranty. Over the past month, I had noticed a rapid decrease in battery life, to the point where I was having to charge it at least once every day. It used to make it almost a week on a charge. I chatted with the AI bot, and it asked me to grant permission for it to run diagnostics on my ring. After I granted those permissions, it said that it detected a problem with my device and was sending me a new one immediately and asked me to verify my address. About a week later, the new ring showed up. It was the most efficient customer service chat experience I’ve ever had (human or otherwise).
I have (had, now) an HP Printer – it’s a little one but does a great job. But when we upgraded our wifi/router and I needed to connect it to the new wifi, I couldn’t do it (there’s no standard screen.) Enter AI – went around in circles – do this, I said it didn’t work, do that, same. Then right back to the first suggestion. It was driving me crazy. Ordered an Epson printer and gave up. What a waste of money (& they lost a future customer. I had only owned Epsons before the HP and now won’t go back.)
Oh, I have so many stories I could tell. But just from this week:
My car dealership chatbot, “Janet,” was not providing a helpful answer. I asked, “Are you human or a bot?” Janet said, “I am a bot, but I can help you.” I answered, “I don’t think you can because so far you are not providing the information I need.” She answered, “I understand. I get that a lot.” (ha, ha)
Second was from a certain national airlines whose website is extremely opaque about its refund and credit policies. (Like, would it kill you to list the about 10 different passenger ticket levels you have and what the policies are for each one?) The chatbot obviously just mines the website because it couldn’t tell me the policy for the ticket I was contemplating, either. I ended up giving up buying a ticket for the day because I was so frustrated.
This was timely. I had an issue with Sam’s Club’s AI. Here’s the story:
I ordered protein shakes and bars to be shipped. They put them in a bag together and mailed them like that. So, as you could guess, the bars got smashed by the protein drinks. I called and got the AI assistant “who can help you with most things but may make a mistake.” Understatement of the year. The AI bot couldn’t even recognize an order number and kept telling me I was putting it in wrong. I wasn’t. If you try to get past it, it takes THREE times of insisting on a person. Then, you get someone you can’t understand or even hear because the noise in the background is so crazy. Oh, they also reshipped me more protein shakes instead of replacing the smashed bars, so I had to do the whole AI thing again. Their error cost them $90. Their AI bot may cost them my business.
One problem is that you can’t even complain about it. There’s nowhere to do so. Maybe X.
I totally support your rant on AI customer service. Now let’s add AI in accounting (specifically a software with initials QB) and let’s add to the rant. The AI in this software tries to be “helpful” by matching payments to their invoices or selecting accounts to charge to and they mess this up on a regular basis. It takes sometimes hours to fix the mess by manually unmatching and then correcting the application to the right customer/vendors. AI is inappropriate where human judgment needs to take place…