441–450 of 563 posts
Earlier quoted context omitted.
Free upside down cake! Sounds like a winner. (I only trust grocery delivery for things in impervious packaging or impossible to screw up).
It wasn't free! And yeah, it wasn't my idea to put a cake into the grocery order. I also don't trust anybody else to pick out fresh fruit & veg for me.
Unless I’m filled with rage.
Earlier quoted context omitted.
I've tried Amazon's chat function and it's put me in an infinite loop on several occasions. I don't know why they didn't have an agent available (it was business hours). I was surprised that a company with Amazon's resources doesn't have a better handle on this, until I remembered that this is the company that doesn't let you say the reason is "wrong size" when returning clothing items. They're not stupid — they're t…
Amazon support is amazing, there’s an option for chat or a receiving a call and always pick call.
More recently, I was lied to repeatedly by agents who told me that they were going to do something, or had done something already. I had to contact support many times, and the issue was only fixed after I was able to escalate to the executive support team, who confirmed I had been lied to repeatedly, and who were able to fix the problem.
Anyone that has ever interacted with a customer service chat bot knows this is pretty stupid. I’m not sure why people (esp. those in the tech community) think we have evolved the technology in any meaningful enough way to replace people yet. How do we know this? Those same people would be against AI chatbots replacing their children’s teachers.
I'm assuming that you're looking at it from the point of view of someone who only contacts tech support when human intervention is actually required. And honestly, probably 90% of people are like this. In terms of who actually takes up the majority of the support reps' time, however, is the 10% of people who are bored, lonely, genuinely mentally handicapped, whingers, and people who get really angry about some bizarr…
I heard this is especially prevalent for suicide hotlines - because of the strict requirements to take and stay on such calls they have to put up with a lot of sexual harassment and its more or less swept under the rug by supervisors.
Earlier quoted context omitted.
> esp. those in the tech community It’s a vocal minority of the tech community that shares that opinion. The silent majority knows it’s just fud marketing. Personally when i encounter a customer support chat bit i immediately assume low quality services and products. When i meet a software engineer blown away but chatgpt’s “coding skills” i assume that engineer is not particularly competent. There are valid use cases…
Few people are "blown away" by the absolute skill level of ChatGPT. It's more that there exists an AI that can do this at all . This just wasn't a thing three years ago, and the first truly capable version (GPT 4) only become publicly accessible this year . Some things I've seen GPT 4 do is "as good as human", often better than average. For example, spotting bugs or reverse-engineering code and then writing doc comme…
This seems to be true regardless of whether humans or robots are involved. If humans are involved they will be low-paid hourly workers on the other side of the planet, staring at the same web page that you are, with no more ability to change anything than you have. And it will take 30 minutes to get one of them on the phone because "we are experiencing unusually high call volume" which translates as "the new vice president fired 10 more customer service agents to make his numbers look good."
If robots are involved it means "the new vice president wanted his numbers to look even better."
Companies have apparently run the numbers and decided it is cheaper to effectively ignore and dissuade dissatisfied customers than to try to resolve their issues. And since most companies insist on binding arbitration now, they're not even worried about class-action lawsuits.
Earlier quoted context omitted.
Communication preference is one thing. But what a lot of modern "online" support is really about is preventing people from getting support and forcing them into a dark pattern doing what the company wants and not what they want. If I could unsubscribe, deal with billing errors, resolve complaints, etc online I'd love it. But most support systems try and trap you in some Kafkaesque faq hell that never lets you get nea…
It's ironic, but at the company where I provide support (50-ish employees), one of the most common things I have to tell people is that they actually have to sign in and cancel their memberships themselves. They just fire off an email rather that investing even four seconds into typing the same message into the Support Center or just looking at their Account page and doing it themselves.
The email flow is click on the "contact us" email that's at the bottom of every page, it opens in gmail, the ask to cancel and hit send. Some time later they get a notification that it's done. They never even had to sign in. They're authenticated by virtue of it being sent from the email associated with the account.
The web flow is signing in, hunting around your app for the account page, and then hunting around that for how to cancel. Sending an email works for every service in the same way. Finding the account management page is different for every website.
Anyone that has ever interacted with a customer service chat bot knows this is pretty stupid. I’m not sure why people (esp. those in the tech community) think we have evolved the technology in any meaningful enough way to replace people yet. How do we know this? Those same people would be against AI chatbots replacing their children’s teachers.
Anyone who has ever interacted with a tier 1 customer service agent in an outsourced support farm knows the difference between that and a chat bot is negligible. True technical support can't be replaced yet, but thousands upon thousands of tier 1 human drones can be.
I took down all the details, and, briefly detailed that she tried everything she could on her end. All previous tickets were closed with "seems fine", basically. Which ticked me off, but I tried to be somewhat professional about it, other than spending more time on that than my employer would have liked. But then I also did what was totally verboten, I wrote down her customer number, and a week later I checked in on her account again. As I expected, the ticket got closed again. So I did the SUPER MEGA verboten thing (because we got paid for inbound calls), I called her, asked if the problem still persists, it did. So I went through all the details again and wrote a super detailed ticket including the medical situation, that she had replaced the phone and everything else she did. I knew better than to be full on snarky, but I also did not hold back. I was fuming. 1-2 weeks later I checked again, it was solved, I called her, she said a tech had shown up and been super nice and helpful. I don't recall what the issue was, but they fixed it. I was so proud, and I couldn't even tell anyone about it.
The only reason I did that was that I figured if they fire me from this minimum wage job I can get just another minimum wage job -- but I would not be able to live with myself to just go by the script in such cases. And I had plenty discussions with team and project lead about my average handling time, you bet. But they never fired me -- because I did solve problems, and with things that were standard procedure I was quick, I ended up quitting.
A bot cannot do that, be disobedient. Too few people can even do that.
Earlier quoted context omitted.
> esp. those in the tech community It’s a vocal minority of the tech community that shares that opinion. The silent majority knows it’s just fud marketing. Personally when i encounter a customer support chat bit i immediately assume low quality services and products. When i meet a software engineer blown away but chatgpt’s “coding skills” i assume that engineer is not particularly competent. There are valid use cases…
Few people are "blown away" by the absolute skill level of ChatGPT. It's more that there exists an AI that can do this at all . This just wasn't a thing three years ago, and the first truly capable version (GPT 4) only become publicly accessible this year . Some things I've seen GPT 4 do is "as good as human", often better than average. For example, spotting bugs or reverse-engineering code and then writing doc comme…
Example: "In JS, create a Promise that is resolved with some emitted value from an EventEmitter, or rejected if an 'error' event is emitted first. Write that concisely, using '.once' and only removing the other event respectively."
Fairly everyday thing in node. Every competent programmer would grasp the intent here pretty quick and write correct code, but somehow it's generally beyond ChatGPT.
Example ChatGPT output (comments mine):
function eventToPromise(emitter, event) {
return new Promise((resolve, reject) => {
emitter.once('error', reject); // memory leak: not removing the other listener
emitter.once(event, (data) => {
emitter.removeListener('error', reject);
resolve(data);
});
});
}
function promiseFromEvent(emitter, eventName) {
return new Promise((resolve, reject) => {
const resolveHandler = (data) => {
emitter.removeListener(eventName, errorHandler); // memory leak: should be removeListener('error', errorHandler)
resolve(data);
};
const errorHandler = (error) => {
emitter.removeListener(eventName, resolveHandler);
reject(error);
};
emitter.once(eventName, resolveHandler);
emitter.once('error', errorHandler);
});
}
ChatGPT: "Apologies for the oversight. You are right; there is a bug in the previous code. The issue is that if the 'error' event is emitted first, both the 'error' and 'data' event listeners should be removed, and the Promise should be rejected."
Nope. The error event listener is actually the only listener working correctly... It proceeds to introduce a cleanup function that removes both listeners (despite keeping '.once'), fixing the bug by accident, not intent. Asking it to change it so that the code adheres to the original prompts starts a downward spiral.
Earlier quoted context omitted.
Communication preference is one thing. But what a lot of modern "online" support is really about is preventing people from getting support and forcing them into a dark pattern doing what the company wants and not what they want. If I could unsubscribe, deal with billing errors, resolve complaints, etc online I'd love it. But most support systems try and trap you in some Kafkaesque faq hell that never lets you get nea…
There's a pretty good trick to dealing with this (I've found): just call up the credit card company and tell them that you tried contacting the merchant, then ask them to bounce the charge
For example I wanted to cancel one of my credit cards. There was nothing in their FAQ's that mentioned anything around "canceling". There was nothing under "Account options" or any logged in setting menu. There was no chat/support on their webpage. They only have a phone line you can call. The phone line is completely automated, and has 2 menus to go back and forth, with no option to speak to a real person. You listen to the first menu, nothing is related to speaking to support or canceling a card. You listen to the second menu, same thing. You go back to the first menu, just to make sure you didn't mishear anything. The call disconnects because you took too long to figure out an option, Eventually you figure out you have to choose one of the options, then spam invalid input.
This finally gets into holding for a real support person. Unfortunately turns out all their agents are busy at the moment and there is a greater than 10 minute hold to get a support agent. If you would like you could hold your place in line and they will call you back when you are ready to take the call. You think to yourself, "sure why not?", and the call ends.
You get a call from their automated system 20 minutes later. It is finally your turn to speak to a human. An automated voice says: "Press 1 to confirm you are ready", you press 1. Nothing happens. The message repeats. You press 1 again. Yet again the message repeats. You spam press 1, the call ends.
You call the automated line again. Go 1 option deep and spam numbers again, get put into holding. This time you do not opt to have your place held in line. Shitty and horrible compressed music blasts into your ear. After about 20 minutes, you get a person. You explain you're canceling, support agent goes through the usual walk through and finally cancels your card.