Notices by Nikolaus Weiskopf (nikweiskopf@social.mpdl.mpg.de)

I am moving from NikWeiskopf@mastodon.online, since the Max Planck Society has their own #Mastodon server now (social.mpdl.mpg.de). Bear with me while I am trying to navigate the process of moving accounts.

@Iris Thank you for the clarification - I get it now. My understanding was that ChatPGT minimizes this risk by design but I am not sure. I thought that this design decision also increased the concern that it could be used without being identified by the standard plagiarism detection methods.

@Iris I am wondering how much stochastic parroting the human brain and cognition does, considering statistical learning models. Is there such a clear cut difference?

If ChatGPT/AI could be enhanced to properly reference prior work/ideas, would it mean that it could be used, since it would not constitute plagiarism any more?

Is #AI comparable to tools like calculators or computers, which we regularly use? Literature search databases? Spell checkers? Symbolic algebra tools like #Mathematica? Similar to translation tools like #DeepL? Where and why would we draw the line? How would we use it and include it in a paper?

I would be interested in opinions, legal studies and other references on this topic, which will certainly keep us in #academia busy for the foreseeable future.

3/3

Maybe I should add: I would be less interested in opinions regarding the question whether you can let #AI write an entire paper but more regarding its use for smaller parts, refining language and restructuring of self-written text? 4/

I think it brings up difficult questions about use of tools in #writing. I have a hard time to follow the argument by #Science that it is simply plagiarism considering its definition as „Plagiarism is the fraudulent representation of another person's language, thoughts, … as one's own original work.“ (https://en.wikipedia.org/wiki/Plagiarism). Is #AI a person when it comes to authorship? On the other hand, #Nature makes the point that #AI cannot be a co-author, since it cannot be accountable.

2/3

Can you use #ChatGPT when writing papers? 🧵

The #scientific #journals #Science and #Nature have developed very different policies for the use of #AI, #LLM, #ChatGPT etc. when writing #papers.

Very simplified:
- #Science regards the use of #AI as plagiarism https://www.science.org/doi/10.1126/science.adg7879
- #Nature allows the use of #AI when acknowledged https://www.nature.com/articles/d41586-023-00107-z

1/3

#introduction

Hi,

I am a #physicist and #neuroscientist who tries to bring both fields together in #neuroimaging.

I direct the Dept of Neurophysics at the #MPI for Human Cognitive and Brain Sciences (https://www.cbs.mpg.de/en). Our main tools are #MRI, #histology and #biophysical #modeling for in-vivo histology (#hMRI) and ultra-high resolution #fMRI.

Excited to see #MastodonScience growing so quickly 😀

This is my personal account and my posts/toots are just my personal views.