Hacker News Reader

When I call a call center, I’m not looking for suggestions on how to solve the problem myself. This builds up nuclear level rage. If I’m calling it’s because I want to delegate the task of solving my problem to someone who can click buttons and make it happen, fast. And if an AI can do that quickly and happily, then good riddance to those call center workers.

Sometimes interacting with these call center workers is so infuriating, that I’d wager this is probably the most hostile interaction an average person probably has with another live human on some regular basis. When I think back to the times when I’ve been really angry in my life, talking to call centers has consistently ranked as the maddest I’ve ever been, and probably the maddest I’ll ever get. Imagine all of that simply going away with a competent AI – a net good for the world.

You get filled with rage because call center tech support is not for you. Call center support rarely is allowed to do anything you can't already do online and they are rarely able to tell you secret information that can't already be found online, and you have already tried online.

Call centers exist to serve the people who can't or won't RTFM online. You do RTFM online, so... rage.

AI-based support is likely to be the same: A chatbot version of the same old information you can already get on the FAQ and Troubleshooting section online. I don't know why we expect it will be different or suddenly unlock some Secret They Dont Want You To Know.

> Call center support rarely is allowed to do anything you can't already do online

This is not my experience at all.

In the past couple weeks, I’ve had to call in to businesses to get the following resolved because it wasn’t something I could do online:

1. Get some bank charges reversed because the online interface explicitly said I wouldn’t get charged for the actions that prompted the charges. 2. Get DRIP enabled on one of my investment accounts. 3. Downgrade a credit card product to a different credit card. 4. Get pricing information about a financial product that wasn’t provided on a banking website. 5. Follow up with the fibre provider in my area because they’ve somehow connected every house on my street but tell me that my house isn’t eligible. (TBF this one isn’t actually resolved yet.) 6. (Twice ) call a bank because legitimate transactions got flagged as fraud and denied. 7. Get my address on file updated with a political party I’d donated to so they’d stop sending mail to an old address.

I can think of at least one scenario where you need that call center worker. Let's say internet goes down in your area, but you're not sure where the problem is. Allowing a customer to call for a technician to come check the connection in your house might be a waste if the problem is regional. If too many people do that, the waste of resources compound. And you could argue that AI could diagnose where the problem is and choose what to do, but maybe it can't. You need that human to disambiguate the course of action.

I don't know what call center you worked in but when I did it, no one cared about any of that, even if they did most customers didn't have the technical acumen to understand what they could have done to avoid the call. They don't want to know how to solve it themselves, they see solving problems as your responsibility since they paid for the product or service and you are it's support. Business customers are the only exception.

Matter of fact, attempting to educate customers instead of telling them what needs to be done to fix the problem can backfire because regardless of your approach it could come off as either condescending or blaming them for calling in.

Self included, I went to my bank a while back to do something and the guy told me I could have done it online and how,instead of just doing it and letting me leave asap. Yeah, I don't give a shit what i could have done online, i already spent too much time fighting their bullshit and I just needed the problem gone asap because it was an urgent situation. At least solve the problem first before you try to educate me.

Call centers have tiers, I do think LLMs can do tier-1 work because tier-1's basically work of a script/flow and LLM would just pretend to listen and react according to the script. Tier-2 and above however is a terrible fit for LLMs.

The reason why the ultimate goal should be to solve the problems before the customer calls in is because of the amount of time wasted by both parties once things get that far. If an AI agent can reduce the time spent by both parties (no waiting on hold for the customer and no employee time for the company), then the goal is accomplished.

For example, you call out documentation as something that needs to be improved so that customers can solve their own problems—what's wrong with that documentation taking the form of an interactive agent you can ask questions to? A well-trained chatbot (not raw GPT) has the potential to be much more useful for the kinds of queries that would otherwise end up at customer support than does a search engine on a bunch of docs.

> what's wrong with that documentation taking the form of an interactive agent

Among many others, in no particular order:

* (borrowing from above) The interactive agent can force you to sit through however many menus/questions the company wants until you are "granted" cancellation. Good luck finding the right option if they make it difficult to navigate! All the while, you have a condescendingly "friendly" "conversation". At least with a real person, I can convey my message clearly, and even if they must still read certain things, they will know to do so more quickly. Have you ever dealt with Amazon's chatbot, which shows the "typing" graphic for a few seconds before giving what is obviously generated text, and it takes a few back-and-forth exchanges like this before someone real introduces themself in the chat?

* You can't quote it like you can actual documentation, which has knock-on effects such being harder to hold the company to account, not being able to give advice to friends & family, not being able to compare competitors before you open an account, ...

None of these problems are inherent to chatbots, nor are any of them unique to them. I'm not even convinced that any of these would be exacerbated by a chatbot—a company that's going to treat its users poorly will do so regardless of the type of agent they employ. And on the flip side, a company that cares about its users could absolutely design a chatbot that solves these problems and legitimately saves everyone time.

I think the attitudes that we have right now towards chatbots are shaped by the companies that don't give a shit—the tech really wasn't ready yet, so only the worst companies tried to force you to use a chatbot. We're on the verge of that changing, and soon a chatbot will be able to provide a better customer support experience than the average overworked call center employee. When that happens, I expect to see companies that care about their customers begin to make the switch, and we'll start interacting with implementations that really work.

Like you, I'd love to live in a world where, "if everyone behaves, things are much better for everyone". It's a great idea.

> and a pathetic, dehumanizing customer retention plea

State your desire to terminate. When a "conversation" begins, simply & robotically repeat your desire. Repeat as necessary.

It works.

We need chatbots on our phone to talk to their chatbots.

I heard that if you encounter one of those "Please tell me what you want help with?" voice prompts on a customer service call, you can just keep on repeating profanity and you will eventually get an operator.

I have never tried it, but I sort of want to...

FedEx hangs up on you. I tried.

Or use a virtual card (such as privacy.com for US). Create one to subscribe and delete it to unsubscribe.