AI/LLM policy discussion

[swansont]

The following was posted in the forum announcements

 

AI-generated content must be clearly marked. Failing to do so will be considered to be plagiarism and posting in bad faith. IOW, you can’t use a chatbot to generate content that we expect a human to have made

Since LLMs do not generally check for veracity, AI content can only be discussed in Speculations. It can’t be used to support an argument in discussions. 

Owing to the propensity for AI to fabricate citations, we strongly encourage links to citations be included as a best practice. Mods and experts might demand these if there are questions about their legitimacy. A fabricated citation is bad-faith posting.

Posters are responsible for any rules violations stemming from posting AI-generated content

 

___

 

We are happy to discuss the whys and wherefores, and consider modifications.

In addition, a reminder that accusing people of being bots, or using AI, is off-topic. You are, however, free to ask for clarification in any discussion, including links to any citations. Faking a cite is easy, but a valid link with one is a little harder to manage.

[Ghideon]

11 hours ago, swansont said:

Since LLMs do not generally check for veracity, AI content can only be discussed in Speculations. It can’t be used to support an argument in discussions. 

Can I assume that AI content can be discussed in the Computer Science section as long as it is used in correct context?

Example:  A member wants do discuss the scientific reasons for choosing between competing AI-based prompting strategies: "In the linked peer reviewed paper the LLM version X was observed to provide an improvement of 3.6% in test Y when using AI generated prompts based to template Z, below is a table of examples." 

The context in the example is science related to LLM's rather than the application of scientific-looking output of an LLM. In this case discussing AI content may be appropriate outside of speculations section?

 

(English not my first language; so the answer may possibly be obvious)

[swansont]

Yes, AI can be discussed. But the arguments one makes in any discussion can’t be AI-generated. AI can’t be used as a source of information.

IOW, you can’t support an argument with anything that’s equivalent to “ChatGPT said <something>” 

[swansont]

On 4/19/2024 at 5:20 PM, swansont said:

We are happy to discuss the whys and wherefores, and consider modifications.

In addition, a reminder that accusing people of being bots, or using AI, is off-topic.

Bumping this. Posts making such accusations with be deposited in the trash can