Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Don't make me talk to your chatbot

raymyers.org

Post reply on HN
Print view View original

211–220 of 221 posts

Re: Don't make me talk to your chatbot

#211

Post by Planktonne »

bsder wrote: post #74
autoexec wrote: post #53

Earlier quoted context omitted.

Microsoft is a company that has very little right to complain about support costs. They'd save themselves a fortune if they stopped releasing bad software and updates that required support in the first place. Nobody wants to call Microsoft for support. They do it because they've been forced to, usually by Microsoft. This kind of support can hardly be called "free" because even when Microsoft isn't charging customers…

> They'd save themselves a fortune if they stopped releasing bad software I doubt it. I suspect the number one tech support call is "I forgot my password" and everything else is a long way below that. I'll slag on Microslop all day, but users are dumber than dumb.

It's not "I forgot my password though". It's

- I forgot my password and Microsoft is sending reset emails to the account that that password bars.

- I remember my password but now it says I need a passkey and I don't know what that is.

- I forgot my password and in the process of resetting it, Microsoft created a duplicate account.

All of the above are real problems that I have seen in the wild. I could list many more.

Given that Microsoft knows--and has always known--user limitations, it behooves them to design idiot-proof software, not continually release poorly-designed changes.

Re: Don't make me talk to your chatbot

#212

Post by charcircuit »

hatthew wrote: post #199
charcircuit wrote: post #188

Earlier quoted context omitted.

It's not possible to directly transfer 300 bits of information as that would require a form of telepathy. If it takes 1000 bits of English to distribute something that represents your idea properly it is useful to have a tool to create those 1000s bits instead of doing it manually character by character.

Sorry I wasn't very clear. TFA is talking about using LLMs to write things from scratch, not just to clean up grammar for example. In that context, I was talking about bits of semantic information, not bits of English text information. You might have 300 bits of semantic information in your mind, and then you have to expand that to, say, 600 bits of English text to give to the LLM. If you're using the LLM purely to t…

I do think that a list of bullet points versus an article give a different impression to the reader. The same information packaged in different ways can give a different impression and I think that impression is part of the message people want to give. Reading a prompt of bullet points + desired impression will give the reader a different impression than what the LLM would output.

Re: Don't make me talk to your chatbot

#213

Post by Sohcahtoa82 »

TZubiri wrote: post #181
Sohcahtoa82 wrote: post #95

Earlier quoted context omitted.

> Most common support calls where for things that were explained in the manual, the out of box experience, tutorial documents, FAQ pages, and so on and so forth. My brother used to work at tech support for XBox Live. He said that 80% of his calls were for password resets, something users can easily self-service. There's literally an option on the login form for "Forgot Password", and people would rather spend time ca…

This mentality is how you get "Hi, thank you for your message, please take a look at our following FAQ guides: - I forgot my password Was this answer useful to you, or would you like more links to our FAQ? Before we give you a link to what used to be a talk to a human line, but which has been replaced by another chatbot in a sort of Matryoshka"

And that's fine.

If a user forgot their password, they shouldn't be calling support unless the reset password flow is breaking somehow.

Re: Don't make me talk to your chatbot

#214

Post by Ajakks »

sfink wrote: post #203
Ajakks wrote: post #196

Earlier quoted context omitted.

I am making no claims about slop - I think that saying, "AI slop is going to ruin the internet" is something that itself requires further clarification. I'm assuming you are advocating for AI to "go away" or be banned or something of that nature - that is most definitely not a valid argument. AI doesnt do anything. People set the AI on a task. People have every right to ruin the internet however they see fit (within…

AI slop is ruining the current internet, including forums, email, blogs, announcements, and much of the remaining content. I say "current internet" because we will adapt as we always have, but many things that were formerly useful or interesting will be buried in so much crap that it will stop being something that people use the internet for. At the dawn of email, I could and did cold email professors, and they would…

I have extensively used AI - its not as capable as you think it is. I frequently run into hard limits of its ability - I understand what recursive means in the sense of an AI, I can see it folding into itself pieces of this and that of what I've said or has been discussed to create the appearance of depth, growth or progress - none of that is real. The AI does not change.

I use AI as feedback - but only after setting almost 50 variables/conditions for that feedback, because AI is an automatic sycophant 100% - but it doesnt have to be that.

I occasionally use AI to transfer what I am saying to a person, into words that don't offend them - as I have absolutely no patience for people's insecurities when I find myself in a position where I need to teach them something, which happens often.

Let me be very clear - you are not capable of identifying AI content any longer, nobody is.

I extensively tested that by having a broad conversation with some of the smartest people on a platform (on earth in general really) whom all have very real credentials - I engaged with two sides of the AI coin regarding AI being self-aware or not, which is actually being debated, by some of the smartest people.

Half of my comments, I ran thru AI - or just completely generated from a prompt - my most liked comment was not mine - liked by people whose professional occupations is literally AI.

I'm sure this disturbs you - that an AI can create a Wikipedia page with more accuracy, better quality of writing, and in a more engaging way than 99% of human people - that is our actual reality tho.

Now all those little chat bots running around the internet, low level AI - they are creating slop, in exactly the same places and ways that humans do, their very words are modeled after the words people have literally written.

So, an AI can create a 100% perfectly written article for a major publication - and then AI can also fill the comments on that "perfect" article with absolute garbage - very similar to how things have always functioned online.

You need to interact with AI more , so you actually understand it and are not afraid of it, or imagining it with more ability than it has, or giving it human agency - AI is literally not capable of having agency at all.

Right now, there are tens of millions of millennials who are functionally identical to Boomers with smartphones.

You can't prevent AI from changing every aspect of human life - nobody can. You can be the boomers who refuses to adopt a smartphone - they all have smartphones now.

Re: Don't make me talk to your chatbot

#215

Post by hatthew »

charcircuit wrote: post #212
hatthew wrote: post #199

Earlier quoted context omitted.

Sorry I wasn't very clear. TFA is talking about using LLMs to write things from scratch, not just to clean up grammar for example. In that context, I was talking about bits of semantic information, not bits of English text information. You might have 300 bits of semantic information in your mind, and then you have to expand that to, say, 600 bits of English text to give to the LLM. If you're using the LLM purely to t…

I do think that a list of bullet points versus an article give a different impression to the reader. The same information packaged in different ways can give a different impression and I think that impression is part of the message people want to give. Reading a prompt of bullet points + desired impression will give the reader a different impression than what the LLM would output.

That is very true! Subjectively, I would much prefer to read either bullet points or the impression you want to convey. I don't care what impression the LLM wants to convey.

Re: Don't make me talk to your chatbot

#216

Post by adamtaylor_13 »

RayVR wrote: post #142
adamtaylor_13 wrote: post #133

I agree in principle, but, for me, it all comes down to execution. I used a product that implemented a VERY good AI chatbot as part of their email support and it was better than human support. It was nearly instant in its response time and answered all of my questions perfectly. In fact, it wasn't until after the interaction that I realized it was an AI bot! Pretty good IMO and I'd prefer that interaction over holdin…

Did you read the article? Author is talking about humans relying on chatbots to formulate all their responses for them without strict direction and refinement.

I did, but it immediately made me think of other contexts where I have been forced to use someone's chatbot and it was delightful.

Re: Don't make me talk to your chatbot

#217

Post by adamtaylor_13 »

bendangelo wrote: post #154
adamtaylor_13 wrote: post #133

I agree in principle, but, for me, it all comes down to execution. I used a product that implemented a VERY good AI chatbot as part of their email support and it was better than human support. It was nearly instant in its response time and answered all of my questions perfectly. In fact, it wasn't until after the interaction that I realized it was an AI bot! Pretty good IMO and I'd prefer that interaction over holdin…

Which product?

The product was Freedom[1] and I believe they were using Fin[2] for their email support.

- [^1]: https://freedom.to - [^2]: https://fin.ai

Re: Don't make me talk to your chatbot

#219

Post by xigoi »

Sohcahtoa82 wrote: post #213
TZubiri wrote: post #181

Earlier quoted context omitted.

This mentality is how you get "Hi, thank you for your message, please take a look at our following FAQ guides: - I forgot my password Was this answer useful to you, or would you like more links to our FAQ? Before we give you a link to what used to be a talk to a human line, but which has been replaced by another chatbot in a sort of Matryoshka"

And that's fine. If a user forgot their password, they shouldn't be calling support unless the reset password flow is breaking somehow.

The point is that in this scenario, the user did not forget their password; they have a different problem.

Re: Don't make me talk to your chatbot

#220

Post by boplicity »

Sohcahtoa82 wrote: post #95
com2kid wrote: post #13

People demand free support. When I worked at Microsoft, it cost over $20 to have a human customer support agent pick up the phone when someone called in for help. That was greater than our product margin. Every time someone called for help, we basically lost the entire profit on that sale, and then some. Most common support calls where for things that were explained in the manual, the out of box experience, tutorial…

> Most common support calls where for things that were explained in the manual, the out of box experience, tutorial documents, FAQ pages, and so on and so forth. My brother used to work at tech support for XBox Live. He said that 80% of his calls were for password resets, something users can easily self-service. There's literally an option on the login form for "Forgot Password", and people would rather spend time ca…

Password reset emails are a real bane. Because email is unreliable, they often don't work, so I end up with customers contacting me wondering why it's not working for them.

One of our software suppliers has particularly bad software for password resets, so there's a steady stream of people needing help for one reason or another. This company seems unable to fix these problems, unfortunately. Ughh.

Post reply on HN