Finally, Some Generative AI Blowback

[Kelley Graham]

If quality matters... Stop Enshitification. From Stackoverflow. https://meta.stackoverflow.com/question ... -is-banned

Temporary policy: Generative AI (e.g., ChatGPT) is banned.

All use of generative AI (e.g., ChatGPT1 and other LLMs) is banned when posting content on Stack Overflow.
This includes "asking" the question to an AI generator then copy-pasting its output as well as using an AI generator to "reword" your answers.
Please see the Help Center article: Why can't I use Artificial Intelligence tools to generate answers?

This is a temporary policy intended to slow down the influx of answers and other content created with ChatGPT and other generative AI technologies, typically using Large Language Models (LLM). What the final policy will be regarding the use of these and other similar tools is something that will need to be discussed with Stack Overflow staff and, quite likely, here on Meta Stack Overflow.

Overall, because the average rate of getting correct answers from ChatGPT and other generative AI technologies is too low, the posting of answers created by ChatGPT and other generative AI technologies is substantially harmful to the site and to users who are asking questions and looking for correct answers.

The primary problem is that while the answers which ChatGPT and other generative AI technologies produce have a high rate of being incorrect, they typically look like the answers might be good and the answers are very easy to produce. There are also many people trying out ChatGPT and other generative AI technologies to create answers, without the expertise or willingness to verify that the answer is correct prior to posting. Because such answers are so easy to produce, a large number of people are posting a lot of answers. The volume of these answers (thousands) and the fact that the answers often require a detailed read by someone with significant subject matter expertise in order to determine that the answer is actually bad has effectively swamped our volunteer-based quality curation infrastructure.

As such, we need to reduce the volume of these posts and we need to be able to deal with the ones which are posted quickly, which means dealing with users, rather than individual posts.

So, for now, the use of ChatGPT or other generative AI technologies to create posts or other content here on Stack Overflow is not permitted. If a user is believed to have used ChatGPT or other generative AI technologies after the posting of this temporary policy, sanctions will be imposed to prevent them from continuing to post such content, even if the posts would otherwise be acceptable.

NOTE: While the above text focuses on answers, because that's where we're experiencing the largest volume of such content, the ban applies to all content on Stack Overflow, except each user's profile content (e.g., your "About me" text).

1. ChatGPT is an Artificial Intelligence based chat bot by OpenAI, which was announced on 2022-11-30. Use of ChatGPT is currently available to the public without monetary payment.

[everything]

how do they stop the AI copy pasta? an "AI" to detect AI?

[Kelley Graham]

the reason stackexchange is so successful as a trusted source is that they have a team dedicated to curating all content. all volunteer. kinda like rsf. ish.

[everything]

right but it's the same issue in education. can humans or their "AIs" detect AI-generated content? so far it seems like the answer is no.

if they are detecting code that doesn't actually work well (chatGPT 3.5 gives me example code that isn't great, but I hear version 4 and other better things give great output) OTOH, that seems like a great idea no matter what. ultimately, that's very temporary.

[BruceP]

the reason stackexchange is so successful as a trusted source is that they have a team dedicated to curating all content. all volunteer. kinda like rsf. ish.

Kinda like RSF, maybe not - ChatGPT being used to attack a poster on a personal level beyond the pale of 'spirited discourse'.

The irony of that particular individual hiding behind the AI to 'examine' the mental state of someone else...defective grey matter swimming in the 'delic and 'tropic soup of false reality, talking about the power of the mind...after crying about being called on their mental illness.

Boundaries for thee, but not for me...

https://rumsoakedfist.org/viewtopic.php ... 19#p528234

[Steve James]

it's the same issue in education. can humans or their "AIs" detect AI-generated content? so far it seems like the answer is no.

Long before ChatGPT, when I was teaching, I'd always have students write their tests in class. It was the only way I knew that what they produced was their own. Students with disabilities were given more time in a proctored space. If it was composition, I could judge their mastery of grammar, mechanics, etc. Otoh, if it were a literature or philosophy class, I could examine their thoughts, which were more important than syntax. I.e., if I could understand what they were trying to say, that was good. If their grammar got in the way so much that I couldn't understand them, that was different.

However, for take home exams, I'd encourage them to use every tool a professional writer might use, including an editor and copywriter. If ChatGPT had existed, I'd have encouraged them to use it as much as wikipedia or the encyclopedia. But, what about students who cheat and just use an AI, or copy from an encyclopedia, or have someone else write an essay for them?

Yeh, that's a problem. They don't want the education that comes from doing the work themselves. My point is that it's not a new problem. Alas, nowadays, many students can only print; so, writing in class is an issue. Imo, I'll take a fifteen minute sit-down with a student to an essay.

[everything]

I remember trying to learn some basic stats by hand, with paper and pencil. doing the little math calculations to calculate means and variances and so on to try to find a line (for linear regression). there is something to be said for doing it the hard way. perhaps it's a trivial example, though.

there is some current hubris from some that says "I understand AI so I can better use it". perhaps that is true for some code generation. I'd rather understand some code to realize if the code I generated is good. but in other (most???) use cases (such as, say, video games), the creators are not the best users.

[Steve James]

That's the thing, you need someone to decide if the answer isn't garbage to us. Though, nowadays, any answer might sound reasonable to someone -who'll get mad if you say it's garbage.

[everything]

the other weird thing is (granted all my code needs are pretty darn easy imho) that I just don't need stack overflow, period.

even with chatGPT-3.5's not-great code examples, it's way more than "good enough" for me to figure out the rest faster than I could.

" this is like rejecting your GPS app in favor of asking random strangers for directions" - what ChatGPT said about this stackoverflow example when I asked it to give me a silly analogy.

[origami_itto]

the other weird thing is (granted all my code needs are pretty darn easy imho) that I just don't need stack overflow, period.

even with chatGPT-3.5's not-great code examples, it's way more than "good enough" for me to figure out the rest faster than I could.

" this is like rejecting your GPS app in favor of asking random strangers for directions" - what ChatGPT said about this stackoverflow example when I asked it to give me a silly analogy.

I usually need it because the problem I'm running into is some weird version incompatibility or bug.

I work on three cloud providers across like 30 different tenancies and they insist on old school system management of the hosts. Ridiculous problems percolate up to me.

[everything]

I don't think I could do sysadmin type stuff. Some AI suggestions would surely help. Way back in the day I did some "webmaster" sort of work of partial Unix admin, partial web dev, partial analytics. Those O'Reilly books were my "ChatGPT". A paper map vs. GPS. Ah the mid to late 90s.

[D_Glenn]

I saw a good meme
“Spend 8 hours writing code then 1 hour debugging.”

“Spend 1 hour using AI to write code, then 8 hours debugging it.”

[origami_itto]

I don't think I could do sysadmin type stuff. Some AI suggestions would surely help. Way back in the day I did some "webmaster" sort of work of partial Unix admin, partial web dev, partial analytics. Those O'Reilly books were my "ChatGPT". A paper map vs. GPS. Ah the mid to late 90s.

It's more automation stuff. The middleware modules change and get incompatible and you have to unearth the one that's screwing you.

I'm a principal admin/level 3 support, usually if a problem gets to my level it's going to require a coding or process change to fix.

I saw a good meme
“Spend 8 hours writing code then 1 hour debugging.”

“Spend 1 hour using AI to write code, then 8 hours debugging it.”

No lie, I was using bing ai to do some Excel-fu, no help at all, wild goose chases

[everything]

Ah you’re a unicorn admin type with some next level patience haha

[Dmitri]

how do they stop the AI copy pasta? an "AI" to detect AI?

That is definitely a huge concern. I'm fairly certain that at some point in the near-enough future an AI-generated output will be completely indistinguishable from that of a real human.
Complete with (simulated) typos, minor grammatical and logical errors, etc.