Login
You're viewing the front-end.social public feed.
  • Chloé Azencottcazencott@lipn.info
    Sep 13, 2026, 9:47 AM

    @mjb @frincon I sadly don't see how I can guarantee none of my coauthors ever use commercial AI tools to help them write (reformulate sentences, translate from French, etc), even though I am not using them myself, so I cannot in good faith commit not to publish any mathematical text generated by AI models (unless my research papers don't count as mathematical texts, which could be argued, but I'm more interested in the spirit of the membership commitment than in its letter). Which is not a happy Sunday thought.

    💬 2🔄 1⭐ 0

Replies

  • Chloé Azencottcazencott@lipn.info
    Sep 13, 2026, 10:14 AM

    @mjb @frincon The little girl who sat next to me on the train platform, looked intently at the blanket I was crocheting, and after a minute or two exclaimed "It's so pretty, what you're making! Who taught you? Can you teach me?" has no idea how much she helped me stop brooding about this.

    💬 0🔄 0⭐ 0
  • Onno (VK6FLAB)vk6flab@mastodon.radio
    Sep 13, 2026, 1:07 PM

    @cazencott @mjb @frincon

    The Debian Linux community decided that their stance would be that the tools didn't matter, but that the contributing person, ie. in your case the author, would ultimately still be responsible.

    While that doesn't address the societal and environmental concerns with Assumed Intelligence and Unsupervised Text Extruders, it does leave a clear person who is required to sign off on their ownership, copyright and licensing.

    You can absolutely say that you don't use it.

    💬 0🔄 0⭐ 0
  • Chloé Azencottcazencott@lipn.info
    Sep 13, 2026, 1:39 PM

    @vk6flab yeah, but Debian contributors are expected to abide by the rules of the Debian community, and take responsibility if they don't. My coauthors are not expected to abide by the rules of a commitment I have signed. @mjb @frincon

    💬 1🔄 0⭐ 0
  • Onno (VK6FLAB)vk6flab@mastodon.radio
    Sep 13, 2026, 7:27 PM

    @cazencott @mjb @frincon

    I understand. Debian has a social contract.

    I would have thought that you have some form of existing binding agreement with coauthors, given that you would likely already be at risk for things like plagiarism.

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Chloé Azencottcazencott@lipn.info
    Sep 14, 2026, 4:36 AM

    @vk6flab @frincon @mjb No, but although I'd be surprised to learn my coauthors are having ChatGPT output entire sections of work, a sentence here and there they had a hard time turning? That's very likely to happen, and not to be perceived as plagiarism by anyone.

    💬 1🔄 0⭐ 0
  • Onno (VK6FLAB)vk6flab@mastodon.radio
    Sep 14, 2026, 5:33 AM

    @cazencott @frincon @mjb

    Perhaps we're miscommunicating.

    As an author you likely had to put systems in place to protect against claims of plagiarism long before chatgpt existed. I'm suggesting that those existing systems might be expanded and used to incorporate use of these new technologies by any co-author.

    💬 0🔄 0⭐ 0
  • Chloé Azencottcazencott@lipn.info
    Sep 14, 2026, 6:38 AM

    @vk6flab @frincon @mjb The only system is trust and common understanding of what constitutes scientific misconduct. As long as my peers don't see every usage of these new tools as scientific misconduct, I can either stop working with them or assume they might use them.

    💬 1🔄 0⭐ 0
  • Michael Baranymjb@mathstodon.xyz
    Sep 15, 2026, 11:28 AM

    @cazencott @vk6flab @frincon I think a lot of mathematicians are really struggling with how to have these conversations with collaborators; collaboration can be a difficult balance for all sorts of reasons and the norm in academic maths is to handle informally and often not explicitly.

    💬 0🔄 0⭐ 0