AI and Editorial: Available Tools, Remarkable Results
Thad McIlroy
New book: "The AI Revolution in Book Publishing: A Concise Guide to Navigating Artificial Intelligence for Writers and Publishers"
In the second webinar of the Book Industry Study Group fall AI season, we’re going to be drilling down on a very specific topic: what is the applicability of the current generation of LLM-enabled AI technology to the editorial challenges faced by authors and publishers when editing books?
The session is specifically not about non-book editorial challenges, although most of what we’ll be covering will apply to all kinds of editing tasks. That’s just the nature of the task of editing.
Let me use that as a segue into a dissection of book editorial challenges.
If you google the simple question “What are the different types of editing that are used for book publishing?” you’ll get a range of answers. Some suggest that there are 3 types, or 4 types, or 5. One listed 10 types.?
Mostly it’s about nomenclature.
There’s broad agreement that the tasks go from the grand to the picayune. At the beginning there’s developmental editing, sometimes called “substantive” editing. Next comes the more detail-oriented line editing, which perhaps encompasses “copyediting”, or sometimes that’s listed as a separate category — there are differences. And similarities.
But there seems to be universal agreement that the final editing stage is “proofreading”, i.e., cleaning up whatever errant errors that may still lie upon the page (as we euphemistically refer to “the screen”).
Others observe a broader range of editing tasks. For example, authors are their own editors, forcing all kinds of fixes to a manuscript before sending it to an agent or a publisher (or publishing it themselves).
Publishers engage in a de facto editorial process when they first encounter a manuscript, thinking about what should be adjusted, and what should be removed. Much later, some conscientious publishers subject a manuscript to fact-checking, and sometimes to plagiarism checking. Are those editorial tasks? What about a “sensitivity edit” (which Microsoft Word labels as “Inclusiveness”) to try to head off complaints about unconscious bias, for example regarding age or disability.
领英推荐
Is creating the index for a book an editorial responsibility? It’s clearly a specialty, one that perhaps falls under editorial. The editing task cannot be easily described.
Within this context we’re going to delve into the increasing role of LLM-enabled AI technology that’s now trying to (further) enable the editorial task.
I’ve identified over twenty available AI-based tools, everything from what’s built into the Microsoft suite of software (most prominently Microsoft Word), and what’s a part of Google’s disparate offerings, from email to Google Docs.
We head next to the also-well-known Grammarly, and the less-well-known ProWritingAid, and the Hemingway Editor. Off to the side are authoring apps, like Sudowrite and Lex, some of which include editing tools, many of which do not. Then there are tools like Trinka, focused on “academic, technical and formal writing.”
Suffice it to say that the editorial function is being tackled from multiple angles. But that’s a subset of some of the broader book-connected language challenges, which include transcription, translation and audiobook creation, all of which are tackled by AI-enabled startups.
Grammer-checking, as you no doubt realize, is not a new software skill. It’s been built into Microsoft Word for eons. Grammarly launched in 2009. There were no large language models back then. Grammar checking was rule-based. The software parsed sentences word by word, and faced all of the limitations of sentences parsed word by word — there wasn’t a broader context. LLM’s have radically changed grammar checking, because they can see the full context of a word, a sentence, a paragraph, an essay, or a whole book, and make corrections within the broader context. It’s a sea change.
Join host Peter Brantley , deep-in-the-editing-weeds Erin Servais , and I, this Wednesday at 1 pm Eastern.
If you can’t make it, the recording will be available afterwards. You can register here:
Publisher, Author, Editor.
3 个月Will editing novels be discussed given their unique challenges? I’d love for a machine to help me with so many tedious novel editing tasks. And there are a lot of errors in novel proofreading making it out into the world. For example, paying $30 for a hardcover novel where a character’s name switches in one instance because the author changed it in a later draft but no one caught the one missed instance is annoying. A machine should say: I see a Nancy who has never appeared before. Thanks.
Books | Publishing | Data Driven | Hands On | Proven Track Record
4 个月Thad McIlroy I wonder if you've done any research on Small Language Model use cases for editorial work in publishing. I ask because so much conventional editorial work on books and related content is deeply informed by three general categories or dynamics: the subject matter, its context, and the author's voice. As you know, these dynamics are most often very specific and nuanced wanting solutions well beyond error corrections and fluency. It seems like the development of "editor friendly" SLM that could be customized by ingestion of specific, permissioned, content for specific use cases would be an interesting direction.