‘… most of the criticism is valid, and if we can edit the right way [my emphasis] and make it a really good read, I will be eternally grateful …’, said an author to me after putting a section of their carefully and diligently copy-edited manuscript through ChatGPT for rating and feedback analysis. True story.

Am I scared of artificial intelligence (AI)?

Short answer: No. I explain my long answer in this blog.

AI is part of our lives

We have to accept this fact. And we have to understand what it means for us in our daily lives, in the lives of those with whom we interact and, significant to this post, in our work as language practitioners. While I can’t speak on behalf of all types of language practitioner, I can speak about AI and copy-editing, to a small extent and in the context of my admittedly limited understanding.

AI is multifaceted

We tend to speak quite broadly about AI and I wonder whether it might seem less of a threat if we were more specific. At an elementary level, this might mean distinguishing between generative AI and non-generative AI. (Subscribe to PEG member Renee Moodie’s newsletter for more on AI specifically.) We use non-generative AI every day – smartphones, social media, PerfectIt, Google Maps/Waze and email spam filters, just to name a few. We don’t seem to fear these. In fact, they’re rather helpful: Waze warns about potholes in the same way PerfectIt queries punctuation inconsistencies, for instance.

It’s the generative AI that seems to have caused the fear and the doubt. Generative AI is an umbrella term for AI that creates ‘new’ content, and it is loved and hated in equal measure. It is seen to generate solutions and might even solve problems. It is considered creative where non-generative AI is considered analytical – not an entirely bad thing, surely?

Keep AI in its rightful place

If we understand who AI benefits and in which ways, we can embrace what we like and use it accordingly. Then we can leave what we don’t like in the hands of those who can benefit. For instance, as copy editors we work on texts that may or may not have been subjected to both generative AI (such as ChatGPT) and non-generative AI (such as Grammarly). While we are not the authors, we are committed to elevating the text to its maximum potential. But an entirely AI-generated text probably can’t be elevated without a complete rewrite. And that’s not the scope of a copy editor, so we needn’t imagine we need to take that on. Send it back or do the best you can with the AI-generated text (rework the -ing words and replace frequently used terms such as ‘robust’ and ‘ever-evolving’!). It doesn’t make for exciting work, but it’s the limit of what we can do when authors have relied on large language models (LLMs) to generate text.

However, if we have been tasked with assisting an author to develop instructional material for an educational project, for instance, we might well use generative AI (LLMs for assignment ideas, test questions or ideas about structuring content, and other generative AI for visual and audio materials). This could save us time and may even support our skills by prompting new thoughts or forgotten methods of working. This use of generative AI could be exciting for those who could benefit directly.

Know what you bring to the table

I have no idea how long it’ll be before we have independently functioning and thinking forms of AI (if ever). For now, we know that these systems are dependent on human input in order to offer any output. What I’m getting at here is what we sometimes can’t hear amid the noise and screams of ‘down with AI’ and ‘AI will take our jobs’: we are the human factor. We are NOT artificial.

It’s the skill and competency of professional language practitioners that ensure written texts are meaningful; that the author’s voice is clear; that the targeted audience gets what they expect. AI can’t offer this.

Generative AI depends on the content it can access. This means it compiles information from what’s out there already and puts it together following sets of rules and patterns that have been pre-programmed (excuse my layperson explanation). AI cannot fathom nuance and style. It is likely to use generic terms and repetitive construction. And it doesn’t always get it right.

Non-generative AI is more straightforward and we are grateful for this. More often than not, this form of AI makes our lives easier. It catches errors in our work or helps us to stay connected to colleagues. It filters out rubbish and can even help with useful aspects of content structure.

AI is not as scary as it seems

AI, both generative and non-generative, is useful when used in the correct context by the correct people. Any form of AI should be used responsibly. Let’s return to the comment at the start of this post from a client of mine. I’d like to share how I handled that situation.

I was initially amused at the words ‘right way’, especially given that they had employed my copy-editing services, presumably on the assumption I knew how to do this correctly. I said I was open to discussing this, but that I would also share my reservations.

I explained the value of author voice, authenticity and the personal nature of their manuscript. I cautioned against any LLM reducing their manuscript to something fairly generic and that there was the risk of their material being accessible to anyone else that the LLM might deem relevant, should a search be made in a related field. I noted that I would not follow ChatGPT’s instructions to re-edit the manuscript, but they were welcome to do so after we had completed our work. My client didn’t challenge my logic and we proceeded as two humans on the copy-editing journey.

It was a productive and valuable working relationship. In fact, we are moving into working on a second manuscript.

But the cherry on the top, and to bring my post to a perfect close, my client decided to put the entire manuscript through Grok (an xAI assistant). One wonders whether they still had doubts or just needed confirmation. But this is what my client said in response to Grok’s positive analysis of the copy-edited manuscript: ‘You may find this [the feedback] interesting. A great compliment to your editing prowess.’ Same true story.

Note: No generative AI was used to write this blog!

1098 words