This is a space to continue discussions about the use of generative artificial intelligence, such as chatGPT, in research.
Slides from the original discussion: ChatGPT Discussion - Google Drive
1 Like
Spinlab Policy As Of 2024-08-29
I am slowing starting to converge on what I feel is the appropriate way to use Generative AI tools such as ChatGPT in a research context.
My starting point is that you should, in no uncertain terms, NOT use it for research in any way. This is likely seen as an extreme position and I am certainly willing to be open minded. We should certainly have several conversations about this to converge on what makes sense.
In the meantime, if you decide to use it for your research in a Spinlab context, then, in the context of Spinlab research, you must disclose your use of ChatGPT and related things. This is to be done by saving the chat transcript and posting it to the ELOG. If you do not disclose your use of ChatGPT-like things within a Spinlab research context, then this is, in my view, unethical and therefore a fireable offense. This policy is not written in stone, but it is now the official policy of the group.
There is likely nuance here, so Pablo, Kyle, and I are preparing a presentation to discuss the various issues from our different perspectives during a Research Discussion, Thursday September 12 at 11:00. I strongly encourage you to attend this discussion.
I strongly encourage you to actively think about the appropriate uses for ChatGPT-like things in a Spinlab research context and to articulate and document your perspective in the ELOG.
It is likely that in 10 years from now, there will be a consensus on the community standards within the scientific community after all of the ethical issues have been hashed out. However, at the moment, this is all new and we are all trying to figure out a reasonable way forward.
FYI, below is the statement in the syllabus for my PHY 810 course for a classroom context:
“You are allowed to use Generative AI such as ChatGPT for any part of the HW but you must abide by three rules: - to be fair to everyone in class (including me and the feedbacker), you can only use the free version, if available, and not any of the paid versions - if you use it, then you must disclose this and include the transcript in the HW submission - any mistakes are your responsibility and will be penalized (in other words, you’ll need to fact check the output and provide appropriate and real citations)”
I am always happy to meet and discuss this face to face, but I strongly encourage you to do your own research on this and to document your perspective first in the form of an ELOG entry.
FAQ
Can I Use Generative AI For Spinlab Research?
Yes, but you must abide by export control laws as well the guidance below from MSU, Funding Agencies, and APS. In addition, you must disclose all use of Generative AI for Spinlab Research.
Can I Use Generative AI Personal Uses or Coursework?
Personal use is up to you, that is outside the scope of our contractual obligations and data management plans to our funding agencies.
For coursework, talk to your instructor for their policies.
How do I disclose my use of Generative AI for Spinlab research?
Briefly describe how you used it and post the full chat transcript to an ELOG entry.
What is unethical about using Generative AI?
This is complicated and depends on context and use case. In my view, it is unethical to not disclose use of Generative AI for Spinlab Research. The Spinlab Research policy is focused on disclosure of use and not currently how about you use it (to the extent that use is consistent with guidance from MSU/Funding Agencies/APS).