Every time I do a search, the bot usually provides enough info. The problem is that I don't trust the bot because it's been wrong before. So I still need to continue the search to find an authoritative answer.

That's the problem with bots. If you trust their answers and they're wrong, it can be a real problem. There was a story a while ago about an Air Canada customer service bot that was giving out bad info about bereavement travel. When a customer tried to get the promised refund, the airline admitted the advice was wrong, but claimed the bot was a separate legal entity, therefore they were not responsible for the advice it gave.

I recently had an important package get rerouted to a pickup box, but the incompetent driver (who did not even manage to reply "hello package" on the intercom) forgot to leave the retrieval card in my mailbox.

After searching the DHL website for a contwct option for way longer than it should take, I saw a small note stating that their chatbot can give you retrieval codes, so I decided to try that.

It was not AI powered, but provided prompts to click, or asked for very specific short replies like the tracking number. Similar to an automated phone system but with text. I took a wrong turn at some point and had to start over, but it actually worked really well and I got the code after about 3 minutes.

They don’t exist to answer your questions. They exist to frustrate you into giving up.

Kinda. I tried activating my new SIM card through the horrible web interface (oh o2 germany never change) an the interface kept bugging out. After a few tries I tried to find the hotline number through the chatbot and lo and behold the chatbot actually scanned my account and deducted that I was trying to activate my new sim card. After a few questions, the chatbot succeeded with what should be easily possible through the web interface.

In my humble opinion we should combine all support bots into one, give it a militarized body, and send it to the front. Nothing will ever get through.

Yes, but not for a good reason.

I’ve asked chat bots for some services (telcos, etc) for very basic questions that you’d think would have been easy to find from a search engine or their main FAQs, but they were not. The bot was at least able to reference some obscure, super old help pages that had otherwise fallen off the radar.

The bot was a solution to a problem only caused because their website was shitty. I guess it’s cheaper to add an ai bot than it is to organize your documentation.

Every time I am forced into that chat box I spam help until I get connected to a human. If thay doesn't work I leave the website and try disconnect the service. I've tried many times to get them to give me a useful answer but they never do.

Shockingly, the bot on my ISP’s website was able to refresh my connection without human intervention, taking care of an issue at the time.

I think we are just a bit biased and disconnected from the general user.

You see, the problem is not that the bot is generally useless. It's just that bots are not 'useful' to you in particular (and to most Lemmy users as of today). Most of the people using this particular website are used to reading a lot of technical (and non-technical) stuff and therefore, are generally good at understanding most computer and internet concepts, which is something I cannot say for the general population outside of these circles. Imagine the old lady trying to buy something on the 'interweb' or people who are just not particularly computer literate or 'tech savy'.

In your case, you mention a bot giving you an answer that you yourself were too lazy to look-up. I'm pretty sure lot's of people don't know where to search and/or don't even bother reading/searching answers to their problems.

See? You just read a whole comment from a stranger because you were curious.

Edit: In reality I agree with you, bots are incredibly frustrating and I do not have any evidence to them ever being helpful, I just wanted to entertain a different idea.

I wish I could turn the amazon ai generated bs in the product info search off permanently, as I only tend to use that search to figure out what actual people have said about the thing I'm looking at and I don't trust a LLM to not hallucinate some random bs for something important I'm trying to figure out thanks.

Maybe once or twice. Mostly I'm trying to get to a human though.

In a useful way? Never.

There was one on the alien site that would respond to a question like "what is XYZ?" with a link to the Wikipedia page for XYZ

Can’t wait to see independent verification of what may be aggressive marketing by Klarna:

Cheaper, faster, better, they say (however they implemented it, presumably spending a metric f🤖kton).

Anybody actually tried it? Klarna specifically, not another company who might be implementing dogshit and/or with devs who have no idea what they’re doing.

No

I know there were a bunch of subreddits with bots that automatically commented Are you looking for this specific thing? on every post, and a lot of posts were answered correctly by those bots.