Facebook Posts
This survey of how AI is affecting the writing and publishing of books came out in today's Publishers Marketplace. Full credit to the author, Katy Hershberger. The link is at the end but there's no way to give a gift link, so I'm excerpting generously (because I think this is an important and very timely set of problems).
Occasionally I run chunks of my own writing through AI-detecting programs, and so far they always come back as "100% human". But how do these programs come to their conclusions? That's entirely opaque. Since AI has been trained on at least 13 of my published books, can't my own prose now come back as AI-generated? Those concerns sit side-by-side with "I don't want AI building on my own work in order to steal the job I love."
**
Academic Publishers Develop Guidelines, Screening Tools for AI Use
Just as booksellers are inundated with AI slop and publishing is reckoning with suspicions of AI-written books, academic publishing is facing similar issues...Submissions to academic journals have ballooned, and, while it’s impossible to prove, many of the papers have the signs of AI, including mistakes and authors with a large number of publications in just a few years, medical journal editor Itamar Ashkenazi said in a recent presentation for the Committee On Publication Ethics (COPE).
Generative AI can be useful in academic publishing, such as for gathering data or translating text from non-English speaking researchers. But the stakes for AI misuse in this world may be even higher than in trade books: Scholarly articles inform scientific inquiry, funding, legislation, medical treatment, and more. When publishers have licensing deals with tech companies, as Wiley and Springer Nature do, their content drives specialized outputs and trains new AI systems. If scholarly content is jeopardized—for example by invented research or fake data—the effects can be far-reaching.
"Each time this flawed information is repeated or used, the error grows, ultimately threatening public knowledge," Chemical & Engineering News states.
Academic publishers of books and journals are beginning to include AI guidelines for submissions, including MIT Press, Oxford University Press, University of California Press, Princeton University Press, and Science journals. There are no universally agreed-upon uses across the industry and some recommendations are more detailed than others. But most publishers agree with COPE’s position, that the author is responsible for the work and they must be transparent about any AI use in their acknowledgements and/or with their editors. The organization has a list of signs of generative AI, including "overly formal and neutral tone," consistency in sentence length, "heavy use of dashes and ‘filler’ words," and "inconsistent formatting."
OUP’s stance for books is stronger than for journals, which just requires disclosure. For books, authors must obtain permission from their editors to use AI in their work and complete an AI Use Declaration Form that explains their reasons for using AI and how they evaluated the validity of its outputs.
"All authors must seek permission if they wish to use AI-generated content as part of their submission and will be asked for it to be replaced with human-generated content, if appropriate," an OUP spokesperson said.
Princeton University Press, which just publishes books, updated their author contract with a clause that requires disclosure of any AI use for creating text or illustrations, director Christie Henry said. Editors are also having proactive conversations with authors about LLM use and discuss in-house what LLM tells to look for, but don't use any machine tools for vetting due to concerns about accuracy.
Henry said the company is "working to refine" their opinion on how much AI use is too much, and came up with categories to describe AI uses:
Assisted - AI helped you brainstorm or research but the writing is yours. Meaning, if you asked AI to find the relevant information for you and then you read those source materials and wrote from them, that's a mechanical assist. It’s just AI working as a search engine.
Collaborated - AI drafted parts while you rewrote, verified, and made it your own. Meaning, if the facts, ideas, etc. came out of the AI model itself (e.g. you asked it explain something to you) and wrote from its answer instead of the source materials, that is AI collaboration and should be disclosed.
Generated & reviewed - AI produced the artifact, but you read every line, checked the facts, and stand behind it. To be clear, you hold responsibility for its integrity and accuracy.
Generated & unreviewed - AI made it and you didn’t really check. This is not acceptable AI use for PUP staff ever: internal, external or under any circumstance.
The last two categories, while informative, are verboten for authors at PUP. Henry notes that the press "would not accept, nor be able to copyright, an AI-generated and reviewed manuscript."
Wiley takes detection one step further, screening every paper not just for AI, but for accuracy and validity. The company proactively checks every submission to their academic journals with a proprietary system called Research Exchange (ReX). The program, which Wiley developed, runs more than 28 checks on a paper’s contributors, content, and context. Each submission is screened multiple times—at submission, after any edits, and after acceptance to the journal. ReX is mostly built on "classic machine learning" according to director, integrity platforms Sam Parker, and uses some generative AI to help explain how it got to a certain determination.
Among the checks on contributors, ReX assesses whether the people are real, have the expertise they claim, if they have retracted papers, and if they submit more than other authors. For content, ReX’s checks include whether a paper’s findings are relevant to the journal, and if a dataset is reused. While it’s not specifically looking for AI use, ReX does search for suspicious words that are hallmarks of LLMs and analyzes a paper’s references—if they are duplicated or hallucinated, it’s a giveaway of AI use. Every report is checked by a human and leads to conversations with the authors. But if a paper is found to have many errors, possibly including AI misuse, it could be retracted.
Wiley’s extensive AI guidelines include best practices for ways the technology might help researchers, noting "Wiley welcomes the thoughtful use of AI Technology" but that "Authors may only use AI Technology as a companion to their writing process, not a replacement." They note that AI-generated content can’t be copyrighted and advise authors, "When possible, use AI for editing and refining your own work rather than generating entirely new content."
...Speakers at COPE’s AI ethics forum noted that it’s difficult and time consuming to analyze submissions for AI use, and that "current tools are not perfect." One suggestion they offer is that "editors may need to insist on having guarantors for each submission as is done for grant applications." For publishers, it all comes down to believing in the work they put into the world...
After the forum, journal editor Ashkenazi replied to additional audience questions in writing. One question indicated that editors at some scholarly journals use at least some level of technology to vet submissions: "How can we expect authors to not use AI to write their manuscripts if we’re using AI to assess their manuscripts?"
"The production of scholarly work by authors and the evaluation of submitted manuscripts by journals are two distinct processes, with different objectives and needs," Ashkenazi responded. "Having said that, both processes currently share the same main challenge. In each process, there are sub-procedures, and it is our duty as an academic/scholarly/professional community to decide in which procedures the use of AI is justified and in which it is not, and in which procedures we will need to declare that we have used AI and in which we will not."
**
Academic Publishers Develop Guidelines, Screening Tools for AI Use - Publishers Lunch
lunch.publishersmarketplace.com
Just as booksellers are inundated with AI slop and publishing is reckoning with suspicions of AI-written books, academic publishing is facing similar issues—and their approaches to the problem, including vetting tools and guidelines for authors, can inform and inspire trade publishers. Submissions...2 days ago
Thank you for doing what you do as a writer and publisher. And thank you for looking into Ai from your perspective. I very much appreciate your input.
My Marketplace account got suspended a few days ago, and despite sending multiple emails to support, I never received any response. I was honestly frustrated and didn’t know what else to do until I came across a recommendation about CyberResolve on Facebook. I decided to give it a try, and thankfully he was able to help me recover my account successfully. I’m really grateful for the help, and I can confidently say my account is back now.
I realize I've primarily seen the results of free public AI use, but I'm still not convinced AI has any place in academic publishing (really, any publishing but especially academic). Just today, I noticed that an article published by one of our local news outlets included an area "map" that was 99.99% wrong---it only got 0.01% correct by having place and highway names that are found in our general area. Guidelines are definitely needed, but I wish we could just put the cat back in the bag. And I want my em dashes back! [My professional background is in academic & patent research so I've watched the evolution of search engines from fully human-controlled to assisted ("assisted" covering a variety of tools, not just AI) to I'm-going-to-try-think-for-you. Even with proprietary databases, I haven't been particularly impressed with where we have arrived.]
Put aside the "how will schools and universities cope with AI?" for a minute, and think about our brains.
I still remember a pivotal class in graduate school: I was doing an M.A. in literature, but I took an elective seminar in linguistics. That class was the hardest I'd worked (possibly, then or since). I had to line up four different schools of linguistic theory and explain what they had in common, and how they were different.
As this was in the Dark Ages, I had to do this without outside help.
I made a chart. I hand-wrote everything that I knew of each school on that chart. I lined up each element of each school side by side. I wrote painstaking paragraphs about each element. I assembled them into an essay.
By the end of it that essay, I *understood* what I was doing. I knew what those linguistic schools were about. More than that, I had learned how to take a set of conflicting, complicated ideas and put them into my own framework.
(And the professor wrote, at the top of the essay, that it was the best one on the subject he'd ever seen. These things matter.)
I've often said that wrestling with a difficult set of ideas and finding understanding is, for me, like the moment when (to use a farm metaphor) you back up the truck with a bumper ball hitch, lower the trailer coupler over it, and feel the sweet moment when the coupler snaps into place. It's a viscerally satifisfying connection. (Maybe some of you can offer a less trailer-based metaphor; I do this a couple of times a week, so that's what comes to mind for me.)
Now, I could ask AI to produce that chart for me. It would be a great chart. It just wouldn't do the same work for my brain.
This week's NYTimes opinion column had an excellent take on this situation.
**
Writing well requires a command of vocabulary, spelling and grammar. It forces writers to comprehend a subject, to sift through their working memory of that topic, to organize and plan. Writing an essay is so difficult that a person in the throes of it can do hardly anything else — as anyone who has shushed a spouse or child while trying to compose a decent sentence can attest.
Very few people would volunteer to dig a ditch. Likewise, it is hard — almost impossible, many English teachers will admit — to turn a reluctant writer into an enthusiastic one.
Yet the American education system expects nearly all students to learn how to write an essay. It’s little wonder, then, that as soon as a tool arrived promising to lessen the pain people grabbed it.
Like it or not, artificial intelligence is now firmly embedded in how students produce text. Somewhere between two-thirds and 90 percent of high school and college students are already using A.I. for schoolwork, according to recent surveys, often requesting help with writing assignments.
A growing number of educators are alarmed, and insist that students should write without A.I. These teachers are returning to handwriting, blue books and proctored, in-class essay tests — all attempts to prevent students from outsourcing their writing to chatbots.
This is baffling to tech industry leaders, who see A.I. rejection as a form of educational malpractice. They point out that writing is already the leading use of A.I. in the workplace, and argue there is no need to continue suffering through the task without help. After all, they say, chatbots now exceed the ability of most human beings to produce fluent prose.
All this points to some questions about writing itself: Why work so hard to learn it? And what do students stand to lose when they stop facing down a blank page, alone with their thoughts?
...Research shows that traditional, unassisted composition is valuable, not despite its difficulty, but because of it. Writing builds our working memory, executive planning skills and metacognition — the ability to think about our thinking, weigh alternate viewpoints and revise for clarity and style. It takes decades to develop writing skills, from toddlerhood into adulthood...
[T]here is no part of the writing process that can be handed over to A.I. without some cost. Intellectual growth occurs at every stage, from brainstorming to fine-tuning a final draft. If writing is thinking, then any part of the struggle that is outsourced to technology amounts to relinquishing some freedom to perceive. How will students ever know what they really think of anything — whether it’s “The Crucible” or A.I. itself — if they ask a machine to spit out an answer for them?
**
Those who want to do the hard work of thinking will write.
Those who don't will cheat.
I'm guessing that there will be an early advantage for the cheaters, and a longer-term advantage for the thinkers. It's possible that our task as responsible parents and pre-AI thinkers is to hold onto some semblance of culture while those advantages shake themselves out.
What Do Students Lose When They Stop Writing? (Gift Article)
www.nytimes.com
Educators are worried that many students can no longer write essays without A.I. And it’s costing more than just grammar skills.4 days ago
This is so small in comparison - I run our church's social media - which is relatively inconsequential. I started using ChatGPT to write the posts for me. It was so quick and I could get a lot of posts done in a very short amount of time. I guided Chat, I gave parameters and asked it to write in the same tone I had been writing in, and it did an ok job of that. In fact, I thought it often wrote posts that were better than mine. However, my creativity took a steep dive. I began to struggle to know what kind of posts to write. I really began to struggle with creating the images - which had been my favorite part. At first, I attributed it to burnout (been doing this for 9 years, now). However, when I started writing them myself again, the creativity has been seeping back in and I enjoy the whole process so much more. I still use Chat every day for some things. I do not use it for my creative pursuits, though.
We need educators to stop stigmatizing, and start understanding, that writing is not the only, or best, way of thinking for everyone.
My students read aloud their essays in class. I had one student read the whole essay and then tell him that it was obviously AI. (It was 98% AI, he had worked on it for 15 minutes the night before.) 1. It was not in his writing voice. Every student has a writing style. 2. It was boring. AI writing is usually boring. It was also long-winded and repetitive. I told him to write an essay on a different subject and he would read it aloud next week, and I wouldn't tell his mom if it was done in time. That one came back as 98% Human and was much more interesting!
This conversation is interesting. I find myself on the other end of the spectrum, learning to use AI to speed up writing technical documents. Learning how to prompt and use different models and then redirecting to get an output that I concur with is mentally taxing. When you speed things up, you are expected to juggle more. The arguments, whether technical or essays, still have to make sense.
I have a very similar memory of taking molecular biology. It was the most difficult class I’d ever taken. We had to synthesize organic chemistry and cellular biology and write cogently in a blue book for our final exam. There was no b.s. Either you knew it or you didn’t haha. But…22 years later I still remember what I learned in that class. I didn’t use it in my career but I’m a better writer and educator for it. Thankfully some schools are pushing back. The classical school my teens attend just issued a strict no A.I. policy on top of an already intentionally low tech focus.
Interesting! I have seen similar arguments for a long time that similarly, mathematical thinking is developed more thoroughly when students are allowed to struggle to an appropriate degree, rather than being spoon-fed each step. I know that when I taught AP Stats, I loved hearing my TA's (former student peer mentors) advise the new crop of students, "Mrs. D. will tell you not to just skip or even just read the examples in the textbook, but to cover up the answers, work through the problem on paper, then compare your response to the longer explanation given in the example. When I first started, I ignore this advice. After the first exam, I started doing it, though, and boy did not only my scores go up, but I really felt like I understood the material." This, and some note-taking practice we did, left me feeling that these kids were heading off to college much better prepared to learn, and to be in charge of their own education, instead of waiting to be spoon-fed information!
I’ve been shocked to see English teachers themselves commenting on articles about AI, saying they encourage their students to use it for brainstorming or outlining, but that the students must do the “actual” writing. 🤯 The brainstorming and outlining, distilling a whole bunch of material and ideas into a coherent whole, is literally the main pedagogical value of a writing assignment! What will become of us? 🤦♀️
Don’t we make these similar tradeoffs with nearly everything we choose to either outsource or do ourselves from scratch? Cooking, yard work, car repairs, childcare. Sometimes our labor bears the best fruit, and other times, the outsource was totally worth it.
Our school district has a no AI policy with the exception of any subject that uses AI as part of its mandated curriculum. I refer constantly to our classroom poster, "Whoever does the thinking, does the learning". No matter, though: I teach middle school Spanish and kiddos still try to use it. They are too naive to realize that I know what they should and should not know!
This is so true. Any kind of writing I do - whether a sermon or a term paper - always feels like a wrestling match - to get the thoughts in my head down on paper into a coherent and persuasive finished product. And you just “know” when it’s not there yet and you also “know” when you’ve won the wrestling match. Truman Capote would work for hours even a whole day on one sentence. No short cuts to clear writing; and definitely not to beautiful, creative, emotionally and intellectually arresting writing.
I have often said that I do not really know what I think until I see what I write about it. Writing is hard, but it is so satisfying, because it is thinking made visible. It is terrifying to me that kids are not learning to write, because in a LOT of ways, it means that they are not going to learn how to THINK. At the end of the day, I love writing, but I do not think it is necessary for everyone to write as much as I do or as well as I do. But it is deeply problematic that so few students are going to learn to do the thinking and the analysis that close writing forces.
I suppose this may change our ability to discuss and debate ideas as well since we aren't "digging deep" into a subject ourselves to really know what we're talking about. As an introvert who prefers deep conversations over shallow ones, I'm concerned.
Have you read the recent piece in "The Atlantic" on Harvard closing their writing center?
It is the journey that forms us not the destination. The extensive writing that I did while earning my degree changed the way my brain worked. I felt sharper and my ability to easily put complex thoughts into language enhanced many aspects of my life. I am concerned for the kids who do not have to take on the arduous task of composing long papers on their own.
I just read the entire article yesterday. Any book recommendation about this topic that would be helpful for homeschooling elementary?
Love this viewpoint. And I agree!!!!
👏
Every skill we outsource to AI with students is a skill we are not training their brains to do. What are we saving all this time and effort for? It seems like we aren’t growing our/their brains in some other direction… but rather just getting lazier. Maybe I’m just old and crotchety as a professor…
