Six-frame cartoon. Frame 1 shows a volcano lair from the outside. Frame 2, implying that the scene is now inside the volcano lair, shows a boardroom table with several editors around it. We see the back of the boss villain editor sitting in a large chair and the other editors facing him. Curly-haired editor says, "Large language models and AI are destroying our livelihoods! Clients don't think they need editors anymore!" Frame 3: We now see the boss villain editor from the front. He is stroking a cat. He says, "We all knew this day would come. It is time." Bespectacled editor says, "You don't mean..." Frame 4: Boss villain editor says, "Release the unnecessary hair-splitting usage distinctions!" Frame 5 shows several rockets, shaped like red pens, launching. Frame 5: Boss villain editor says, "Now go forth and enforce!" while Curly-haired editor and bespectacled editor both rise from their seats to do his bidding.Creative Commons License

This cartoon was inspired by a Twitter conversation with Jonathon Owen. Propose your own hair-splitting usage distinctions using this blank and post them to social media! For example, here’s one of Jonathon’s.

In all seriousness, I’m not actually concerned that AI will leave us jobless. As other editors, including Adrienne Montgomerie and Hazel Bird, have written, AI is just another tool we may have to learn and incorporate into our workflow, but large language models, including tools like ChatGPT, are nothing more than sophisticated text predictors, and they can’t do the important synthesizing and structuring work that human editors are so good at.

In May, I was part of an international panel for the Institute of Professional Editors (IPEd) conference on futureproofing the editorial profession. The panel featured IPEd’s Ruth Davies, Professional Editors’ Guild’s Alexis Grewan, and the Chartered Institute of Editing and Proofreading’s Janet MacMillan, and one of the topics that came up in our discussion was the effect of large language models on editing and editors.

As I mentioned during the panel discussion, different editors will grapple with these innovations in different ways, but I expect to use my editorial skills to further equity. Because large language models are trained on existing writing, which, in English, is heavily biased in favour of the work of men of European ancestry, editors will play a key role in redressing that imbalance by choosing to work with more authors from historically underrepresented backgrounds and bringing their writing to the fore.

I’ll also be devoting much of my attention to fact checking, since, as Audrey McClellan has noted, ChatGPT shamelessly makes things up all the time. My favourite example so far of this phenomenon is courtesy of the lawyer who submitted to a judge an AI-generated brief that cited nonexistent cases.

Finally, I’ll bolster the user-testing side of my plain language services. The newly published ISO plain language standard uses the International Plain Language Federation’s definition of plain language, which centres the intended audience as the only arbiter of plainness—meaning that user testing is a non-negotiable part of the plain language process. Beyond user testing, editors can also explore different models of co-creating with audiences, which could open the door to more relational ways of producing written work and to new genres we haven’t imagined yet.

4 thoughts on “Conspiracy”

  1. I’m embarrassed to ask, but … how do you add text to the image?

    For the record, my favorite hair-splitting distinction is the one I learned as a baby copyeditor: a pompom is a weapon; a pompon is a fuzzy decorative ball.

  2. I know that this was just rhetorical flourish, but “shamelessly makes things up” sorta-kinda is suggestive of the very thing that can confuse people, since ChatGPT et al. have little agency and no sentience: AI can’t (well, as of now and as far as we know) have any emotions about its work.

    PS am I being That Guy, Mr Well-Actually? Apologies if so.

    1. Fair criticism!

      But if AI is incapable of feeling shame, then it’s literally shameless, no? 😉

Leave a Reply

Your email address will not be published. Required fields are marked *