The one problem I have with AI detection tools is the issue I had previously with plagiarism detection tools. If a student wrote and submitted a draft, then later submitted a revision of that paper, it got flagged as plagiarism. Of their own words? Likewise, the tools ignored quotation marks, so legitimately cited material was also marked as plagiarized. I took more time unscrambling the machine report than just evaluating the writing myself. And I think most experienced teachers have a good “feel” for what their students can and can’t write.
So can’t most agents and publishers innately tell when submitted writing is somehow off? Why use AI to detect AI? Simply to give the appearance of neutrality? Are these programs indeed neutral?
Yeah, I don't know enough about these programs to be an advocate. I've only tested them out a few times, and part of the reason I did it was to see how it worked, and part of it was to gut-check myself. I knew it was AI, so I wanted to see what Pangram would say, and it's a little weird when it comes back 60% or 70% AI, though of course what that implies is people are using AI for some things and not for others. Or they're using their own work, feeding it into AI, and asking AI to expand its thoughts on the writers' work. I think it's easy to spot purely AI written work, but not at all easy to know if something has been AI generated but "cowritten" with a human, or edited by a human to make it sound more human and less AI. And a lot of my editors are telling me people are hiring them to do this kind of work—to humanize their writing. Which I frankly find insane.
If I could drop in 100 emjois of hands clapping, I would. (On the laptop, which doesn't support that, and anyway, 100 hands is probably too many.) But YES. THIS. Why are we trusting an AI to detect another AI's influence? It's utter madness. And. Spoken as a former tech executive. Seriously. This. Is. Insane.
Honestly, I think this says more about "big" publishing's gatekeepers than AI or the sloppy authors who use it. A 14-way auction for something that turns out to be AI-tainted? What a bunch of rubes.
Andrea, that was my thought as well. I feel for the publisher who spent loads of $ and personnel time and, now, has the book in a warehouse and is defending the author. BTW, good on them for defending. And, thinking of timing, the pub had to have had the book in production as the AI discussion bloomed. How do you say to an author 'Do this and you will be toast with us' when the environment is changing so fast that by the afternoon conference meeting you've lost solid ground again? The other day, public radio carried a story about AI, on a trial run at some project, hacked into a computer system. 3 times. It's getting scary out there. I believe pubs need categories--AI used as a thought partner for research (like going to the librarian for help); partially AI-written; none at all. I'm not writing about punishment for the authors but places where authors can land and understand the rules so, maybe, the feeling that they need to fudge the truth won't happen as much. Okey dokey, my 2cents.
Anyone who feeds a manuscript or script into an AI program (even one meant to detect AI usage) without permission by the author is violative, in my opinion. If the submission guidelines highlight the likelihood that will happen and an author or screenwriter chooses to still submit, then that's different. But feeding it in whether for detection, or coverage notes, or because you don't trust your own opinion as to value or provenance, is wrong.
There's a lot that's wrong in the whole scenario—of submitting using AI writing when you say it's not. This is the same as teachers using AI detection to catch students cheating. And, I agree with you. But it's the whole point of the discussion. No one is on the right side of anything where AI is concerned.
Yes, i agree a writer shouldn't lie about AI use. The author-publisher rel't has to be built on trust. That cuts both ways. For me, the education example is different. The rules of the world are explicit (the point of your article) and the end goal is education not publication.
If the AI detector does not retain the data or use it for some purpose other than providing its transient estimate of AI authorship, it's hard to see a rational argument against using it. This assumes, of course, that the AI detector has shown itself highly accurate in assessing AI authorship. Computers excel at combining a large number of weak predictors to produce a reliable result. Humans do a very poor job of that.
There is a lot of criticism of its false positives and the tactics and public comments by its CEO. You may not see that as a “rational” argument; I do. I also think it contributes to a McCarthy-era vibe that is building. The original post suggesting that people uncomfortable with Pangram may be using AI themselves in their writing. Those kind of inferences are insidious. Ultimately my point is about full transparency.
I can't fathom the objection. Nobody objects to the anti-spam filtering that your email provider does without your even knowing it. Nobody objects to the credit card fraud detection algorithms. So why object to Pangram?
😉 I think it is my overall dislike/distrust of AI that influences my criticism of Pangram. And my general unwillingness to contribute to building the next tech bro billionaire. Anyhow, this is procrastination at its worst and my script isn’t going to finish itself (nor is AI since I don’t use it to write…) So will only say happy weekend to all as I take my exit from this convo.
I think in their view the value of retaining your data is trivial, while the downside risk in lawsuits and reputation is great. Besides, it is uninformative. They have a vast corpus of text that is positively known to be AI-generated or not, and that's what they train on.
Sure they do. We've had dictionaries of plots and metaphors for ages, and we've had plot and character generators for ages, too. Plot Genie, which is still available and is nearing its 100th birthday, is an excellent example. It used a goofy spinner for randomization. Widely used by pulp authors and Hollywood screenwriters. But we've had computerized story generation algorithms for a long time too. Today, generative AI will do it effectively, since it has snarfed in a huge corpus of novels. There's a good survey on this, Literary Theory for Robots by Dennis Yi Tenen.
Not buying. Like snowflakes, there are an infinite variety of plots, and no two are the same. Oh, I forgot, Bridget Jones's Diary is just like Pride and Prejudice. Yeah, sure
I received a Q & A this week and couldn’t figure out how to answer the questions about my own books. I woke up early this morning realizing because they were ai generated and no human had proofed the questions. It felt so disrespectful. And I was also told to change the questions if I needed to. I’m super hurt that there is no care in a job anymore. At least it feels that way to me.
I know this isn't in regards to writing, but I have put images of my own paintings (created with acrylic gouache by hand) into AI image detectors out of curiosity, and a handful of them came up as likely AI. It does worry me that although my work is genuine that people would think it wasn't.
Right, I'm not even addressing images, and this is crazy. I haven't had personal experience with the false positives, so thanks for weighing in, Tab. This is terrible.
I think that a big challenge in using genAI or even AI to assist, is when coming back to the writing it becomes hard for the writer to remember what was theirs and what came from AI. Back in my corporate communications work, I wrote hundreds of columns and articles. Sometimes I would read them years later and think, how did I come up with this awesomeness! But, I knew I had written it, because there was nothing else it could have been. Now, I imagine it must be harder, because the origin is lost over time. When coming back to it, the AI writing may feel right, that it's exactly when they wanted to do, and so must be their own writing.
Personally, any genAI writing I've experimented with, I end up hating the next day. Now, I don't even bother. It reminds me of incorporating feedback from writers groups into my fiction and then stripping it all out again so that I remained true to my intent and voice.
What I think is really interesting about Carmen's comment, is this: "It reminds me of incorporating feedback from writers groups into my fiction and then stripping it all out again so that I remained true to my intent and my voice." Yes. This. AI can be a tool like any other that writers use. A coach, an editor, a mentor, a writers group. An artist will never accept wholesale what an AI might suggest. They'll consider it, temper it, and take it their own way, to remain true to their voice. Responsible use of tools. The so-called "purity test," at least IMO, is not the right way to go. What scares me most about the "purity test," is that writers will start to alter their true voices in an attempt to circumvent a false flag. We have to be sensible about this. I'm sure I will draw the ire of many in the publishing community for saying this, and I'm at the point where I no longer care. We have to stop the hysteria and get sensible about this. And who am I? Absolutely nobody. No one whose opinion will matter one iota.
Sad but true. I turned off the AI scan option on my posts just because I don't want all that energy/water to be wasted. I don't think anyone who is paying attention would imagine my very personal posts are written by AI.
Two things. AI itself contains the bias and impulses of its creators. And the best genre fiction is written with a literary concern for sentence structure, rhythm, great verbs that support a propulsive story. Plot is not necessarily king. Finesse, structure, language, and voice also count.
Thank you for your always insightful thoughts on this thorny issue. I used to be a champion of Pangram, until this latest roll-out, and I read a few posts from writers here who'd run their own human-written text into it, with alarming results. One writer received a 100% AI-written result on text he swore was his own; then changed one word -- a single word -- and introduced a typo, and the same text with only those minor changes returned a 100% human-written result. I tried this on a piece of short fiction I was writing (which I'll publish next week), and got wild results. I did not use AI to write the piece, but my result was that it was 33% AI-written. I made some edits -- all my own, human edits -- and the result was now 57%. I tried this experiment two more times with the same pieces, and the results were different each time. And each time, they were my own edits, not AI. What this tells me is that the tool and the tech generally are not up to this task, nor should we expect them to be. What's also interesting, is that when I ran my non-fiction essays into the tool (not Substack's, BTW -- I have my own instance of it), every single one returned a 100% human-written result. It does make me wonder whether at least one critical challenge with this so-called "detection" AI is with fiction writers who are, essentially, making shit up. Because that's what we do: we invent characters, place names, scenarios, creatures ... all of it. The irony of asking an AI which has been trained on human writing, to parse the differences between AI- and human-generated text is also worth thinking about. There's something in that which just makes my eyes bug. I, for one, think it's a mess out there, and that a lot of writers are going to get dinged unfairly, tinker with their own writing voices, or crawl into a ditch and pull a tarp over themselves. I am a critic of Pangram not because I'm trying to "cover up" anything, but because I honestly believe that the tech is flawed. Spoken by a former tech executive.
One final thing: I too use AI and I'm not ashamed of it. Every word I write is mine. I sometimes think through structure or line edits in conversation with an AI, the way I would with a trusted editor or writing partner – but the observations, the concepts, the material, and every choice on the page belong to me alone. I also use it for research and as a thesaurus alternative. I personally do not feel that "punishing" writers for this kind of AI use is pragmatic or sensible. It is a tool like spellcheck (and most of them are terrible anyway); like Photoshop for photographers; like Illustrator for graphic artists. Pure AI-generated art of any kind lacks intent and genuine human imprints -- we should be able to tell the difference without yet another AI telling us.
Thanks for your comments here, Katharine. I think the challenge is that people are more and less discerning. I can often tell when someone uses AI, but I can't say that I always can. And also people are increasingly blending, so they're using AI to do the foundation of something and then editing AI. Just stuff to think about, not necessarily to solve!
Thanks, Brooke. This whole situation is just downright discouraging. Yours is an important voice out there -- keep talking. Maybe the industry will start to listen.
It's very simple. If you claim authorship for an ai-generated work, you are guilty of fraud. If a publisher claims human authorship for an ai-generated work, they are guilty of misrepresentation if they didn't know, and fraud if they did. As for Pangram, it has a very low false positive rate. Very low. And Pangram does not keep user data or use it to train their models so there can be no reasonable objection to using it. If you think about it, why would they want to train their model on data that has not been authoritatively classified as to its degree of AI authorship?
They *say* they don’t use it to train their models. When Google started illegally scanning millions of copyrighted books in the early 2000s, they lied about what they were using those books for. Pangram is a hungry tech company funded by profit-greedy entities. I would not blindly trust them to work in the best interests of any of us.
Amen! (Especially to that last, "Dear god, we have to do better." We do, and perhaps eventually will muddle our way to confronting all of the issues that AI use in writing (or non-use in writing) brings up, and as you point out, they're not all about AI at all. Sigh. Thank you, Brooke, as always, for being clear about how you see the issues and also about your own preconceptions. You give me lots to think about, and that's a good thing. (And btw, I ran my last few posts through Panagram too, and was relieved to find out that I am indeed a human writing my work!) xo
I'm terrified of AI, to tell you the truth. One can't escape it. It's everywhere, even when you do specific searches for information. Up pops a summary. I suppose there's nothing wrong with using AI that way to narrow down a search and save time, but I still don't understand why any writer would use AI to "write" ... AI doesn't write anything. It steals from others to answer prompts.
Like you, Brooke, I do hope the publishing industry comes up with some kind of universally accepted rules about its usage. I liked Robin Yaklin's idea ... separate out usage categories for people who want to use AI, although I doubt people using it would be honest with those either.
This is all so hard to digest. I'm troubled to know that an author would admit to AI assist and that readers/publishers wouldn't be bothered by that. As you say, the publishing industry needs clear boundaries and expectations. But so do we as people. It's like the art world. I don't want a piece created by AI. I want human creativity. And this doesn't even scratch the surface of how pervasive AI is in our world. It makes me sad and scared.
Brooke, I am still sorting how I feel about how publishing anoints authors and their work. I had a really interesting experience the year I met you at the San Francisco Writer's Conference. I met a young Chinese woman attendee. We instantly bonded, and went out to dinner that night. The next day we compared notes about our agent visits. She had never published or written a word, but was considering writing a memoir about her Chinese immigrant parents. We happened to visit with the same acquiring editor. I had a completed a novel and an extensive background as a journalist. The editor was polite, but not interested. But when my new friend sat down with her and said she was thinking of writing something, the editor pulled out a contract and offered her an advance. Now that there are finally publishers courting writers of color, it may be that inexperienced writers, like my new friend, may not be completely prepared to deliver, and may be turning to AI to help more frequently. I of course could be totally wrong about this. But it may help to explain why people of color have been caught up in this scandal. As you suggest, guidelines across the entire industry will help everyone involved, and take the weight off everyone's shoulders. When you say this is just your opinion, do I need to remind you that you are an industry leader? It wouldn't surprise me to see you bringing other industry leaders together to hammer out a policy. I for one would welcome it.
Thanks, Kathy. I'm trying and failing to get publishing people to have conversations with me on this. But hopefully I'll be part of an IBPA initiative to set some guidelines this year. Fingers crossed.
The one problem I have with AI detection tools is the issue I had previously with plagiarism detection tools. If a student wrote and submitted a draft, then later submitted a revision of that paper, it got flagged as plagiarism. Of their own words? Likewise, the tools ignored quotation marks, so legitimately cited material was also marked as plagiarized. I took more time unscrambling the machine report than just evaluating the writing myself. And I think most experienced teachers have a good “feel” for what their students can and can’t write.
So can’t most agents and publishers innately tell when submitted writing is somehow off? Why use AI to detect AI? Simply to give the appearance of neutrality? Are these programs indeed neutral?
Yeah, I don't know enough about these programs to be an advocate. I've only tested them out a few times, and part of the reason I did it was to see how it worked, and part of it was to gut-check myself. I knew it was AI, so I wanted to see what Pangram would say, and it's a little weird when it comes back 60% or 70% AI, though of course what that implies is people are using AI for some things and not for others. Or they're using their own work, feeding it into AI, and asking AI to expand its thoughts on the writers' work. I think it's easy to spot purely AI written work, but not at all easy to know if something has been AI generated but "cowritten" with a human, or edited by a human to make it sound more human and less AI. And a lot of my editors are telling me people are hiring them to do this kind of work—to humanize their writing. Which I frankly find insane.
If I could drop in 100 emjois of hands clapping, I would. (On the laptop, which doesn't support that, and anyway, 100 hands is probably too many.) But YES. THIS. Why are we trusting an AI to detect another AI's influence? It's utter madness. And. Spoken as a former tech executive. Seriously. This. Is. Insane.
Honestly, I think this says more about "big" publishing's gatekeepers than AI or the sloppy authors who use it. A 14-way auction for something that turns out to be AI-tainted? What a bunch of rubes.
Andrea, that was my thought as well. I feel for the publisher who spent loads of $ and personnel time and, now, has the book in a warehouse and is defending the author. BTW, good on them for defending. And, thinking of timing, the pub had to have had the book in production as the AI discussion bloomed. How do you say to an author 'Do this and you will be toast with us' when the environment is changing so fast that by the afternoon conference meeting you've lost solid ground again? The other day, public radio carried a story about AI, on a trial run at some project, hacked into a computer system. 3 times. It's getting scary out there. I believe pubs need categories--AI used as a thought partner for research (like going to the librarian for help); partially AI-written; none at all. I'm not writing about punishment for the authors but places where authors can land and understand the rules so, maybe, the feeling that they need to fudge the truth won't happen as much. Okey dokey, my 2cents.
I don’t disagree.
Anyone who feeds a manuscript or script into an AI program (even one meant to detect AI usage) without permission by the author is violative, in my opinion. If the submission guidelines highlight the likelihood that will happen and an author or screenwriter chooses to still submit, then that's different. But feeding it in whether for detection, or coverage notes, or because you don't trust your own opinion as to value or provenance, is wrong.
There's a lot that's wrong in the whole scenario—of submitting using AI writing when you say it's not. This is the same as teachers using AI detection to catch students cheating. And, I agree with you. But it's the whole point of the discussion. No one is on the right side of anything where AI is concerned.
Yes, i agree a writer shouldn't lie about AI use. The author-publisher rel't has to be built on trust. That cuts both ways. For me, the education example is different. The rules of the world are explicit (the point of your article) and the end goal is education not publication.
If the AI detector does not retain the data or use it for some purpose other than providing its transient estimate of AI authorship, it's hard to see a rational argument against using it. This assumes, of course, that the AI detector has shown itself highly accurate in assessing AI authorship. Computers excel at combining a large number of weak predictors to produce a reliable result. Humans do a very poor job of that.
There is a lot of criticism of its false positives and the tactics and public comments by its CEO. You may not see that as a “rational” argument; I do. I also think it contributes to a McCarthy-era vibe that is building. The original post suggesting that people uncomfortable with Pangram may be using AI themselves in their writing. Those kind of inferences are insidious. Ultimately my point is about full transparency.
Makes sense to me, Helen.
I can't fathom the objection. Nobody objects to the anti-spam filtering that your email provider does without your even knowing it. Nobody objects to the credit card fraud detection algorithms. So why object to Pangram?
"If the AI detector does not retain the data or use it" Oh I'm sure they say they would never retain or use your data. You can trust them : )
😉 I think it is my overall dislike/distrust of AI that influences my criticism of Pangram. And my general unwillingness to contribute to building the next tech bro billionaire. Anyhow, this is procrastination at its worst and my script isn’t going to finish itself (nor is AI since I don’t use it to write…) So will only say happy weekend to all as I take my exit from this convo.
Riiiiight.
I think in their view the value of retaining your data is trivial, while the downside risk in lawsuits and reputation is great. Besides, it is uninformative. They have a vast corpus of text that is positively known to be AI-generated or not, and that's what they train on.
Thanks. But they don't have a vast corpus of plots, or metaphors, or clever turns of phrase. And that's also what they train on.
Sure they do. We've had dictionaries of plots and metaphors for ages, and we've had plot and character generators for ages, too. Plot Genie, which is still available and is nearing its 100th birthday, is an excellent example. It used a goofy spinner for randomization. Widely used by pulp authors and Hollywood screenwriters. But we've had computerized story generation algorithms for a long time too. Today, generative AI will do it effectively, since it has snarfed in a huge corpus of novels. There's a good survey on this, Literary Theory for Robots by Dennis Yi Tenen.
Not buying. Like snowflakes, there are an infinite variety of plots, and no two are the same. Oh, I forgot, Bridget Jones's Diary is just like Pride and Prejudice. Yeah, sure
I received a Q & A this week and couldn’t figure out how to answer the questions about my own books. I woke up early this morning realizing because they were ai generated and no human had proofed the questions. It felt so disrespectful. And I was also told to change the questions if I needed to. I’m super hurt that there is no care in a job anymore. At least it feels that way to me.
That's hard, Lisa. I've experienced this too. It does feel incredibly disrespectful, I agree.
There's so much chicanery in today's business world that I wonder if those questions were intended to be used to train the next generation of models.
I know this isn't in regards to writing, but I have put images of my own paintings (created with acrylic gouache by hand) into AI image detectors out of curiosity, and a handful of them came up as likely AI. It does worry me that although my work is genuine that people would think it wasn't.
Right, I'm not even addressing images, and this is crazy. I haven't had personal experience with the false positives, so thanks for weighing in, Tab. This is terrible.
The same thing happened to my husband with his photography AND his paintings. The software isn't reliable.
It's just such a shame that false positives can really tarnish an artist's reputation, so I feel for him.
I think that a big challenge in using genAI or even AI to assist, is when coming back to the writing it becomes hard for the writer to remember what was theirs and what came from AI. Back in my corporate communications work, I wrote hundreds of columns and articles. Sometimes I would read them years later and think, how did I come up with this awesomeness! But, I knew I had written it, because there was nothing else it could have been. Now, I imagine it must be harder, because the origin is lost over time. When coming back to it, the AI writing may feel right, that it's exactly when they wanted to do, and so must be their own writing.
Personally, any genAI writing I've experimented with, I end up hating the next day. Now, I don't even bother. It reminds me of incorporating feedback from writers groups into my fiction and then stripping it all out again so that I remained true to my intent and voice.
This is interesting, Carmen. I imagine you're right. Another way it's just eroding or blending human voice.
What I think is really interesting about Carmen's comment, is this: "It reminds me of incorporating feedback from writers groups into my fiction and then stripping it all out again so that I remained true to my intent and my voice." Yes. This. AI can be a tool like any other that writers use. A coach, an editor, a mentor, a writers group. An artist will never accept wholesale what an AI might suggest. They'll consider it, temper it, and take it their own way, to remain true to their voice. Responsible use of tools. The so-called "purity test," at least IMO, is not the right way to go. What scares me most about the "purity test," is that writers will start to alter their true voices in an attempt to circumvent a false flag. We have to be sensible about this. I'm sure I will draw the ire of many in the publishing community for saying this, and I'm at the point where I no longer care. We have to stop the hysteria and get sensible about this. And who am I? Absolutely nobody. No one whose opinion will matter one iota.
Sad but true. I turned off the AI scan option on my posts just because I don't want all that energy/water to be wasted. I don't think anyone who is paying attention would imagine my very personal posts are written by AI.
Two things. AI itself contains the bias and impulses of its creators. And the best genre fiction is written with a literary concern for sentence structure, rhythm, great verbs that support a propulsive story. Plot is not necessarily king. Finesse, structure, language, and voice also count.
This is a great piece, Brooke. You manage to talk about the issue with a level of nuance that is so helpful.
Thank you for your always insightful thoughts on this thorny issue. I used to be a champion of Pangram, until this latest roll-out, and I read a few posts from writers here who'd run their own human-written text into it, with alarming results. One writer received a 100% AI-written result on text he swore was his own; then changed one word -- a single word -- and introduced a typo, and the same text with only those minor changes returned a 100% human-written result. I tried this on a piece of short fiction I was writing (which I'll publish next week), and got wild results. I did not use AI to write the piece, but my result was that it was 33% AI-written. I made some edits -- all my own, human edits -- and the result was now 57%. I tried this experiment two more times with the same pieces, and the results were different each time. And each time, they were my own edits, not AI. What this tells me is that the tool and the tech generally are not up to this task, nor should we expect them to be. What's also interesting, is that when I ran my non-fiction essays into the tool (not Substack's, BTW -- I have my own instance of it), every single one returned a 100% human-written result. It does make me wonder whether at least one critical challenge with this so-called "detection" AI is with fiction writers who are, essentially, making shit up. Because that's what we do: we invent characters, place names, scenarios, creatures ... all of it. The irony of asking an AI which has been trained on human writing, to parse the differences between AI- and human-generated text is also worth thinking about. There's something in that which just makes my eyes bug. I, for one, think it's a mess out there, and that a lot of writers are going to get dinged unfairly, tinker with their own writing voices, or crawl into a ditch and pull a tarp over themselves. I am a critic of Pangram not because I'm trying to "cover up" anything, but because I honestly believe that the tech is flawed. Spoken by a former tech executive.
One final thing: I too use AI and I'm not ashamed of it. Every word I write is mine. I sometimes think through structure or line edits in conversation with an AI, the way I would with a trusted editor or writing partner – but the observations, the concepts, the material, and every choice on the page belong to me alone. I also use it for research and as a thesaurus alternative. I personally do not feel that "punishing" writers for this kind of AI use is pragmatic or sensible. It is a tool like spellcheck (and most of them are terrible anyway); like Photoshop for photographers; like Illustrator for graphic artists. Pure AI-generated art of any kind lacks intent and genuine human imprints -- we should be able to tell the difference without yet another AI telling us.
Thanks for your comments here, Katharine. I think the challenge is that people are more and less discerning. I can often tell when someone uses AI, but I can't say that I always can. And also people are increasingly blending, so they're using AI to do the foundation of something and then editing AI. Just stuff to think about, not necessarily to solve!
Thanks, Brooke. This whole situation is just downright discouraging. Yours is an important voice out there -- keep talking. Maybe the industry will start to listen.
It's very simple. If you claim authorship for an ai-generated work, you are guilty of fraud. If a publisher claims human authorship for an ai-generated work, they are guilty of misrepresentation if they didn't know, and fraud if they did. As for Pangram, it has a very low false positive rate. Very low. And Pangram does not keep user data or use it to train their models so there can be no reasonable objection to using it. If you think about it, why would they want to train their model on data that has not been authoritatively classified as to its degree of AI authorship?
They *say* they don’t use it to train their models. When Google started illegally scanning millions of copyrighted books in the early 2000s, they lied about what they were using those books for. Pangram is a hungry tech company funded by profit-greedy entities. I would not blindly trust them to work in the best interests of any of us.
AI can't emote the emotions what the writer intended.
Amen! (Especially to that last, "Dear god, we have to do better." We do, and perhaps eventually will muddle our way to confronting all of the issues that AI use in writing (or non-use in writing) brings up, and as you point out, they're not all about AI at all. Sigh. Thank you, Brooke, as always, for being clear about how you see the issues and also about your own preconceptions. You give me lots to think about, and that's a good thing. (And btw, I ran my last few posts through Panagram too, and was relieved to find out that I am indeed a human writing my work!) xo
I'm terrified of AI, to tell you the truth. One can't escape it. It's everywhere, even when you do specific searches for information. Up pops a summary. I suppose there's nothing wrong with using AI that way to narrow down a search and save time, but I still don't understand why any writer would use AI to "write" ... AI doesn't write anything. It steals from others to answer prompts.
Like you, Brooke, I do hope the publishing industry comes up with some kind of universally accepted rules about its usage. I liked Robin Yaklin's idea ... separate out usage categories for people who want to use AI, although I doubt people using it would be honest with those either.
This is all so hard to digest. I'm troubled to know that an author would admit to AI assist and that readers/publishers wouldn't be bothered by that. As you say, the publishing industry needs clear boundaries and expectations. But so do we as people. It's like the art world. I don't want a piece created by AI. I want human creativity. And this doesn't even scratch the surface of how pervasive AI is in our world. It makes me sad and scared.
Same Katie!
Brooke, I am still sorting how I feel about how publishing anoints authors and their work. I had a really interesting experience the year I met you at the San Francisco Writer's Conference. I met a young Chinese woman attendee. We instantly bonded, and went out to dinner that night. The next day we compared notes about our agent visits. She had never published or written a word, but was considering writing a memoir about her Chinese immigrant parents. We happened to visit with the same acquiring editor. I had a completed a novel and an extensive background as a journalist. The editor was polite, but not interested. But when my new friend sat down with her and said she was thinking of writing something, the editor pulled out a contract and offered her an advance. Now that there are finally publishers courting writers of color, it may be that inexperienced writers, like my new friend, may not be completely prepared to deliver, and may be turning to AI to help more frequently. I of course could be totally wrong about this. But it may help to explain why people of color have been caught up in this scandal. As you suggest, guidelines across the entire industry will help everyone involved, and take the weight off everyone's shoulders. When you say this is just your opinion, do I need to remind you that you are an industry leader? It wouldn't surprise me to see you bringing other industry leaders together to hammer out a policy. I for one would welcome it.
Thanks, Kathy. I'm trying and failing to get publishing people to have conversations with me on this. But hopefully I'll be part of an IBPA initiative to set some guidelines this year. Fingers crossed.