(C) Daily Kos This story was originally published by Daily Kos and is unaltered. . . . . . . . . . . DK: Is it OK to call out diaries that appear to be composed by, or mostly by, AI? [1] ['This Content Is Not Subject To Review Daily Kos Staff Prior To Publication.', 'Backgroundurl Avatar_Large', 'Nickname', 'Joined', 'Created_At', 'Story Count', 'N_Stories', 'Comment Count', 'N_Comments', 'Popular Tags'] Date: 2023-03-18 On the Internet, no one knows you're a...what? (DK library, uploaded by BMScott) To be clear, I don't mean those diaries written (presumably) by humans about AI, quoting AI compositions, such as this diary on Thursday that drew extensive discussion: I asked ChatGPT4 to explain climate change in the language of Shakespeare. The question is about diaries that show earmarks of artificial composition. I've seen a few around here that read that way, per the intuitions of this former professional writer/editor. For example, one fairly recent piece here appeared to me like a student essay or rather slack Wikipedia entry, full of generalities. It did not get many Recs but drew a few favorable comments and some discussion. I initially wrote a substantial comment on that diary, stating the several reasons that I thought the diary was probably an AI composition. There was an author comment, however, which included one small IMO "human" clarifying sentence. I was afraid of potentially hurting a person's feelings by telling them they write like a bot. I cleared my comment rather than post it. In another case, my thought process followed a similar route. I didn't challenge the author about appearing to have used AI, but I engaged in the comments, getting responses that seemed rather human. Later I began to feel guilty about a reflection that might have been harsh, and wrote an apology, which drew a further response that seemed...odd. Later again, thinking....Did I just get punked into worrying about hurting a bot's feelings ? (And is its handler chuckling?) DK Rules of the Road DO #1 is pretty well baked into our conditioned reflexes here (most of us who are very active anyway): DO Remember that there are real people on the other end of your communication.... But what if that is not always the case? *** OK, why not treat AI composition like any other? Just presume there is a real human being on the other end, no harm, no problem? Well, actually, there are several potential problems. As DK member Torta commented on the Shakespeare diary referenced above, although AI didn't nearly match the quality of the real thing, The bot whizzed thru its capabilities and spat that out very, very quickly (I imagine). I’m thinking about the combo of passing-grade competence (in a world where, often, nothing better than that is required) and superhuman speed. So if bots are simply accepted: --Bot postings have the potential to constitute a new and less detectable kind of spam, pushing actual community writers rapidly off the front page, discouraging them, and diluting actual human engagement. --Quality will suffer. Audiences notice, if only subliminally. This would not be good for the site. --Bot postings and/or comments on divisive subjects could easily be engineered repeatedly to absorb people's energy and disrupt the community. I mean more than humans do anyway on our own! In other words, automated trolling. --Bot postings can also be just dreck, or incorporate dreck. To quote from a thoughty (though by no means anti-AI) opinion piece yesterday in The Guardian: We know that there are some real extra risks in using generative AI. It has “hallucinations” where it makes things up. It sometimes creates harmful content. Such dreck, however, takes up human attention and requires human energy to keep under some sort of control--if that is even possible. Do we want to expend resources on that? Will we have a choice? Finally, --A "community" incorporating robots as contributors or commenters with equal rights and privileges seems inherently problematical. * * * Another reason I've hesitated calling out when I seem to hear the hum of machinery back of a posting: DK Rules of the Road DO NOT #13: DO NOT Publicly accuse other users of being paid campaign shills, sockpuppets, zombies, or imposters. Such accusations should be brought up in a private ticket at the Help Desk, with proper evidence. Would or should this also apply to suspected AI bots? We know that the Help Desk has resources to detect or verify certain kinds of bad behavior. But is there any screening mechanism to detect bot prose? In the possession of the Help Desk, or anywhere at all? Surely if there is not something, something will be invented. Every weapon (and AI composition can be used as such, though it has other potentials) eventually generates defenses. Anyone know about this? But probably not yet, as educators are admittedly wrestling with the situation. And meanwhile? Here? * * * I wrote a fairly substantial diary comment, criticizing the AI version of Shakeapeare on climate change. Pointing out--among other things--bad "Elizabethan" grammar. And then read this, in the Guardian piece: [W]e can see [IT's] flaws when we try out these prototypes that the technology companies have made publicly available. You can have a lot of fun getting it to write poems or songs or create surreal images. Ask it a straight question, and you usually get a sensible safe answer. Ask it a stupid or complex question, and it will struggle. A lot of tech experts and journalists have had fun testing it to destruction and making it respond in bizarre and disturbing ways. The AI boffins will be delighted because this all helps refine their programming. They are conducting their experimentation partly in public. [Emphasis added.] Was I merely enlisted into unwittingly helping some "boffin"? :-\ Maybe not this time, but it could happen. Is that what the site should be for? As for AI "harvesting" of DK content for "training"purposes, I suppose it must be already happening. (Would we have any way to know or prevent that?) * * * What if: DK authors had to show human bonafides of some sort on request ? (What kind of bonafides? Whose request?) What if: AI or AI-assisted postings were required to be labeled as such? (As food labels have to note artificial flavors?) What if: DK site copyright information were to note explicitly that content harvesting from the site for AI purposes is not specifically authorized but, even if postings are copyrighted, such harvesting cannot be controlled, detected, or prevented by site administrators? Just some thoughts. I'll close with one more quote from the Guardian opinion piece by Charlie Beckett: It is too late to put this technology “back in the box”. It has too much potential for helping humans meet the global challenges we face. It is vital that we have an open debate about the ethical, economic, political and social impact of all forms of AI. Including, IMO, its potential impact on DK. What are you thinking? [END] --- [1] Url: https://www.dailykos.com/stories/2023/3/18/2158911/-DK-Is-it-OK-to-call-out-diaries-that-appear-to-be-composed-by-or-mostly-by-AI Published and (C) by Daily Kos Content appears here under this condition or license: Site content may be used for any purpose without permission unless otherwise specified. via Magical.Fish Gopher News Feeds: gopher://magical.fish/1/feeds/news/dailykos/