The DMZ

« first day (4452 days earlier) ← previous day next day → last day (726 days later) »

Rory Alsop

09:24

@schroeder it makes up citations and references, in a very convincing way

4 hours later…

doppelgreener

13:00

Chatgpt is not a knowledge generator or a reliable reference

Expecting people to check it and calling people a fool if they don't is a bit far. Its purpose is not to synthesize knowledge, its purpose is to output convincing sentences — at any level of truth.

Actually understanding anything it's talking about is not a property of its function, just a property it appears to present emergent from how it plays word association.

Journeyman Geek

Its plausible garbage. :D

Which is the problem

@schroeder tbh, On 2 of the 3 sites I'm mod on it gets you a year off

Deletion is giving too much of a benefit of the doubt

And I've had ONE person apologise and promise to do better

doppelgreener

If I had a device with a couple of random wheels to spin to complete the sentence "have you [verbed] the [noun]?" With technical jargon it'll be right sometimes, but that doesn't make it intelligent and inherently worth consulting.

@JourneymanGeek yeah I'll just as likely mislead myself entirely

Journeyman Geek

@doppelgreener its the attitude really

"I'll just tell people the stuff from the magic black box so they think I'm smart"

schroeder

13:36

@doppelgreener you're a fool if you don't check Wiki. And like wiki, it might not be correct, but it provides the associated topics, terms, and concepts to further your research.

so just like it is important to check the 2nd page of google results in your search, you need to check wiki, and now that it exists, you need to check chatgpt. Because if you miss something that chatgpt offers, you've missed out.

Journeyman Geek

@schroeder if the point is "give correct answers" sure

if its "low effort rep" or "lets see what happens" ...

schroeder

@JourneymanGeek oh, I'm not talking about SE rep or answers, that's another problem. I'm talking about knowledge management and research skills in general. Which SE needs to adapt to as new sources of "knowledge" emerge

I'm banning questions of the type "chatgpt says X. Is it true?" But I really suspect that we'll need to flex that as the years go by

Journeyman Geek

@schroeder ah

those are also lazy attempts at cheap rep :D

doppelgreener

14:27

i mean, it's a good idea to check a wiki because it's a good idea to check encyclopeias in general. the entire purpose is it gathers together knowledge on a topic. if i want knowledge on a topic, i open an encyclopedia.

chatgpt has a purpose entirely unrelated to this.

it may put some words together that appear in proximity ... but i could just check the wiki and learn that. and it's not like gpt has any trustworthiness in putting those together. if i'm relying on it to learn what words to use or how to use them, i'm setting myself up to be misinformed.

chatgpt has no actual merit here, and has significant negative merit.

doppelgreener

14:41

instead, i can go places where I will be informed, like an encyclopedia

Fire Quacker

@JourneymanGeek Wow, you're way stricter than SO. Back in December when I was flagging stuff, accounts usually got 1 week - 1 month.

Journeyman Geek

@FireQuacker Honestly - it started with MSE in our case

and uh

I was annoyed

and frankly there's little value add from these users and these users are .... kinda trying their luck

so I think applying the FAFO principle here isn't too bad an idea :D

Fire Quacker

Mmm... Yeah, I can see MSE being more strict... If you know enough about the network to post on MSE, then you probably know what things are frowned upon and you're probably doing it intentionally

Yeah, SO was mostly new users or accounts that hadn't posted in years - much more likely to not know about the rules (aside from the banners everywhere)

I have no idea what anybody would hope to gain from using ChatGPT on MSE

Journeyman Geek

@FireQuacker interestingly...

mostly new users

@FireQuacker well ONE was being a smartass

and posting an answer on the question about chatGPT

Pets is generally 'softer' moderation so I just ~~barked at them~~ gave them a fairly stern warning

4 hours later…

Rory Alsop