Formally Ban AI in Forum Posts

han614698

It's just a huge cheater way of doing it, and doesn't help people.

It's already really frowned upon, but I doubt it's reportable.

ilovestories

Is it that much of a problem?

han614698

ilovestories wrote:
(#2)
Is it that much of a problem?

Yes - no one really wants AI in response to their question,

Za-Chary

ilovestories wrote:
Is it that much of a problem?

Yes. When someone asks a question in the forums, it is presumably because they are looking for advice from a human Scratcher. If they wanted an AI answer, they would have just asked an AI. The AI-generated posts also frequently contain misinformation, which oftentimes is subtle to the point where even the Scratcher posting the AI-generated content doesn't realize it's misinformation.

As for the suggestion at hand, I believe this is already implemented: https://scratch-mit-edu.ezproxyberklee.flo.org/discuss/post/7795869/

cheddargirl wrote:
Outside cases of a person is using AI for post assistance, use of AI would not okay for the following reasons:
1) A person looking for answers in the discussion boards are looking for help from a person. If someone wanted to get an AI-generated answers, there's no point they would be asking here instead of seeking help directly from ChatGPT (associated with Microsoft), Bard/Gemini(associated with Google), or some other AI service.

2) The answers posted by AI can sometimes be inaccurate or topic-derailing, and spreading misinformation or posting derailing spam is not okay. It's in violation of the “Give helpful feedback” guideline (formerly the “Be constructive” guidelines.

It's my understanding that you can use the Report button on posts that use AI to answer a question.

han614698

Za-Chary wrote:
(#4)
ilovestories wrote:
Is it that much of a problem?
Yes. When someone asks a question in the forums, it is presumably because they are looking for advice from a human Scratcher. If they wanted an AI answer, they would have just asked an AI. The AI-generated posts also frequently contain misinformation, which oftentimes is subtle to the point where even the Scratcher posting the AI-generated content doesn't realize it's misinformation.

As for the suggestion at hand, I believe this is already implemented: https://scratch-mit-edu.ezproxyberklee.flo.org/discuss/post/7795869/
cheddargirl wrote:
Outside cases of a person is using AI for post assistance, use of AI would not okay for the following reasons:
1) A person looking for answers in the discussion boards are looking for help from a person. If someone wanted to get an AI-generated answers, there's no point they would be asking here instead of seeking help directly from ChatGPT (associated with Microsoft), Bard/Gemini(associated with Google), or some other AI service.

2) The answers posted by AI can sometimes be inaccurate or topic-derailing, and spreading misinformation or posting derailing spam is not okay. It's in violation of the “Give helpful feedback” guideline (formerly the “Be constructive” guidelines.
It's my understanding that you can use the Report button on posts that use AI to answer a question.

That question was, however, asking if they copied the entire thing. The posts from a specific user I've been seeing are three paragraphs (obviously) by AI, then one sentence that says thanks for asking a question.

Za-Chary

han614698 wrote:
That question was, however, asking if they copied the entire thing. The posts from a specific user I've been seeing are three paragraphs (obviously) by AI, then one sentence that says thanks for asking a question.

I would argue that the example you describe here is not considered “using AI for post assistance” that @cheddargirl described.

han614698

Za-Chary wrote:
(#6)
han614698 wrote:
That question was, however, asking if they copied the entire thing. The posts from a specific user I've been seeing are three paragraphs (obviously) by AI, then one sentence that says thanks for asking a question.
I would argue that the example you describe here is not considered “using AI for post assistance” that @cheddargirl described.

What would you consider in the realm of “using AI for post assistance”?

That may be the only reason for this suggestion to stay up - to make the ban stricter. Otherwise I'll have no problem reporting this post to be closed as well as reporting a whole bunch of other posts.

Za-Chary

han614698 wrote:
What would you consider in the realm of “using AI for post assistance”?

If a majority of the post appears to be written by a human.

han614698

Za-Chary wrote:
(#8)
han614698 wrote:
What would you consider in the realm of “using AI for post assistance”?
If a majority of the post appears to be written by a human.

Ok
So would this be allowed?

I used AI to write a post about why we shouldn't be using AI to work with suggestions on Scratch:

Hey Scratchers!

I've been thinking a lot about the potential of AI to improve Scratch. While I'm excited about the possibilities, I'm also a bit worried about the potential downsides. One of the biggest concerns I have is the idea of using AI to generate suggestions for the Scratch website.

Here's why I think AI might not be the best solution for suggestions:

Losing the Human Touch: Scratch is a community-driven platform. Suggestions from real Scratchers are valuable because they come from people who actually use the site and understand its strengths and weaknesses. AI suggestions might lack the same level of insight and empathy.
The Risk of Over-Engineering: AI could potentially suggest changes that are too complex or technical for the average Scratcher to understand or implement. This could make Scratch feel less accessible and user-friendly.
The Danger of Automation: If we rely too heavily on AI suggestions, we might lose sight of the importance of human creativity and innovation. Scratch is a place where people come to experiment and try new things. We don't want to stifle that creativity with a machine-generated list of "best practices."

That being said, I do think there are ways to use AI responsibly to improve Scratch. For example, we could use AI to analyze user data and identify trends or patterns that could inform future improvements. We could also use AI to help moderate the community and prevent harmful content from being shared.

What do you guys think? Should we embrace AI as a tool for improving Scratch, or should we be cautious about its potential drawbacks? Let's discuss!

Happy Scratching!

[Your Scratch Username]

Now, theoretically, would it be okay if I used those reasons listed there in an actual post?

cambegonefr

I sometimes used AI back when I was actually active on Scratch on something to go off of if I didn't know how to word a response correctly. I'd say don't ban it completely but ban certain usages of it.

8to16

Making posts solely to use AI is already banned, but not using AI in posts intended to be used for something normal.

50_scratch_tabs

ChatGPT wrote:
As an AI language model, I see the value in diverse perspectives, but I understand the concern about AI contributions feeling less authentic. While I can provide information quickly, nothing compares to the creativity and insight of real people. It might be worth discussing how we can maintain a balance that prioritizes human interaction while still leveraging technology when appropriate.

Seriously though, I support because even though I like AI, it is honestly not right very often. If nothing else, the ST should require you to disclose if you use AI to generate a response. It can be misleading, and as ChatGPT said, AI just doesn't feel quite right.

starlightsparker

Banning AI in forum posts is counterproductive. AI enhances communication by clarifying complex topics and making discussions more accessible, especially for those with writing challenges. It can also inspire creativity and innovation, helping users generate new ideas. Instead of a ban, we should establish community guidelines for responsible AI use, promoting transparency and quality contributions. Rejecting AI overlooks its potential benefits and limits the richness of discussions. Embracing AI responsibly can lead to a more inclusive and engaging forum environment, fostering diverse perspectives and deeper learning. Let’s focus on integration rather than restriction. -ChatGPT (This is a joke, I do not agree with this block of text.)

Would using AI to create summaries of long posts be prohibited as well?
Would using AI for assistance with BBcode formatting be prohibited as well?
Would using AI to grammar-check (But not to write entire forum posts) forum posts be prohibited as well?

ThisIsTemp1

starlightsparker wrote:
Banning AI in forum posts is counterproductive. AI enhances communication by clarifying complex topics and making discussions more accessible, especially for those with writing challenges. It can also inspire creativity and innovation, helping users generate new ideas. Instead of a ban, we should establish community guidelines for responsible AI use, promoting transparency and quality contributions. Rejecting AI overlooks its potential benefits and limits the richness of discussions. Embracing AI responsibly can lead to a more inclusive and engaging forum environment, fostering diverse perspectives and deeper learning. Let’s focus on integration rather than restriction. -ChatGPT (This is a joke, I do not agree with this block of text.)

Would using AI to create summaries of long posts be prohibited as well?
Would using AI for assistance with BBcode formatting be prohibited as well?
Would using AI to grammar-check (But not to write entire forum posts) forum posts be prohibited as well?

I think the OP is suggesting to ban using AI for your entire forum post, not for post help.

50_scratch_tabs

starlightsparker wrote:
(#13)
Would using AI to create summaries of long posts be prohibited as well?
Would using AI for assistance with BBcode formatting be prohibited as well?
Would using AI to grammar-check (But not to write entire forum posts) forum posts be prohibited as well?

I think summaries would be fine, as long as it's disclosed that it's AI.
BBcode help would be fine, as long as the user checks it and makes sure for good etiquette (like making sure that ChatGPT closed all it's tags).
Grammer-check would be fine, especially if you are replying to a post in another language, especially using Google Translate.

Few hours later and I just realized I was ninja'd

Last edited by 50_scratch_tabs (Oct. 6, 2024 03:26:35)

ThisIsTemp1

I wholeheartedly support. It is disappointing to me that something I do for free, voluntary, and use some of my time I could be doing something else, is being done using AI.

Za-Chary

han614698 wrote:
I used AI to write a post about why we shouldn't be using AI to work with suggestions on Scratch:

[…]

Now, theoretically, would it be okay if I used those reasons listed there in an actual post?

If you use the AI response to write the post in your own words, that should be fine. For example, copy/pasting the quote you gave would not be okay, but writing something like this should be fine:

Suggestions from Scratchers (not AI) are useful because they come from people who actually use the site. AI suggestions do not necessarily have the same understanding of Scratch that a Scratcher does. In this way, AI could also potentially suggest changes that are too complex for Scratch. We don't really need to fill up the Suggestions forum with suggestions that are known to be too complex for Scratch.

Note that some aspects of my quoted post above are directly copy/pasted from the AI's post — others are simple rewordings — and the last sentence is essentially a new idea. The rest of the AI's post didn't really make much sense to me; this is either because I misunderstood the context in which such a post would be used, or because the AI stated wrong information in its post. Finally, the quoted post is not nearly as long as the AI's post, and doesn't read like an automated response, in part due to the lack of unnecessary greeting/closing. This could be an example of a way that AI helps with post assistance without just directly copy/pasting an AI response.

starlightsparker

AI is not a Scratcher, and so therefore AI does not have the same understanding that us humans do of Scratch.

If one asks AI if naming browser extensions is allowed, it will say it generally is, but in reality, it is not. AI does not have this information, as it likely sees what is in the Community Guidelines, and purely bases it off of them. Human forumers, however, would likely have the correct answer as they do not have one source.

And so therefore—I believe creating posts purely with AI should be disallowed in the forums.

Last edited by starlightsparker (Oct. 5, 2024 16:59:46)

TheAutocorrectingCat

I've seen the forumer that OP is talking about. Some forumers use AI for their entire post. This should not be allowed, as AI is not knowledgeable about the way Scratch works. Support.

GlitchedThrough

I gave used AI to reword some statements I've made before, but that's with the initial post being in my own words, and only partially being rewritten by AI. As such, I believe an outright ban may be a bit much, but entire posts aren't great.
However, how would the ST know if it's written by AI?

Discuss Scratch