So, I was reading the latest issue of INCITE magazine, that came out today, and in Jane Cowell's introduction piece, she makes mention of how her team has been exploring ChatGPT in writing media releases, blog posts, and social media posts... and asks the question, "Have you experimented with this technology? Don't forget to blog about it and share your experience with us all."
Which seems as good a prompt as any to get semi-back on track with BlogJune.
Now, I fancy myself as a decent copywriter. About 80% of the time, if I'm given a brief to write something - such as a media release or briefing notes, I can sit down and bang one out a draft within an hour, depending on how divided my time is in that moment. But many people I've worked with struggle at this - as do I that other 20% of the time. You can call it writer's block, or waiting for the inspiration to strike - either way, it's not the best use of work time.
In fact, in some workplaces, I've relied on having a team member write up a first draft for me. Once I have that first draft, my editor instincts kick in, and I have a work that I can pull apart, mould and rewrite, adding tone of voice, nuance and in-house style to arrive at something that better resembles the final draft that I can get signed off.
But that first draft... getting it done is rarely the best use of anybody's time. So why not use an AI program such as ChatGPT to be your artificial copywriting minion?
Another example - I'm working on a marketing campaign, and I need to get a designer to create some visual collateral to be the brand of the campaign. Now I can spend the time writing up a marketing brief, send it to the graphic designer, wait a few days, get a first mock-up, and send it to and fro before eventually coming up with a finished image, maybe up to a week later. Or, I can experiment with an AI image creator, and come up with some mock-up images within a couple of hours. They won't be perfect - at the moment, much of it will end up in uncanny valley territory - but it's something that you can send to the designer and say, "Here's a rough visual indication of what I want. Please create a polished version." Theoretically, it saves so much time in the design process, using AI to create that first mock-up before you even need to talk to your designer.
The thing is, though, AI isn't magic. It doesn't pull that image - or first draft - out of thin air. It creates a derivative work, based on an immense source of textual or visual data which is, on the most part, somebody else's creative work. Presumably without their permission. In fact, there have been well-documented examples of artistic works that AI has blatantly used as the basis for artificially generated artwork.
As a copywriter - or as an artist - you wouldn't just take somebody else's work and adapt it to pass off as your own. There are both moral and legal implications in doing so. So at what point does AI-generated work become plagiarism / breach of copyright? More importantly, as the volume of AI-generated work starts to overtake the volume of unique human-created work, what does that say about the progression of human expression and knowledge?
Honestly, I do think that AI has its place in the work of content creation - but as a starting point, not as an ends in itself. Ultimately, AI can get you part of the way there in putting together the foundation and bare bones, but human expression and creativity is still needed to finish the job well.
More importantly, putting my librarian hat on for a second, we still need to consider how we manage the rights of creative works that we make publicly available in the digital world. We may have been too keen to optimise digital access, and an AI bot isn't necessarily going to pay attention to that copyright statement or Creative Commons license - it just scrapes everything publicly available on the internet.
This may be a case of trying to close the gate after the horse has bolted, but perhaps what we need is some kind of "ethical AI" alternative - one that only uses content / data that has a Creative Commons license that allows derivative works. That would be a start. Or using library image content where the copyright status is clearly and accurately described in the catalogue record. Either way, it's our responsibility as the custodians of such collections to ensure that there is clear and sufficient data that allows for AI to use these collections ethically.