A few weeks ago, I had dinner with some friends. One of them pulled out his phone and asked me if I thought I had written more than 200 pieces or so at this stage of my career. “Probably,” I said. “Why?”
“You know, the way things are going, artificial intelligence is going to be able to write articles and books just by analyzing your writing style.” I was initially incredulous, maybe a bit dismissive. Then my friend went to a program on his phone and put in a query, asking it to write a 1,000-word Christmas story in the style of Charles Dickens. To my amazement, the program started churning out paragraphs within seconds — and not just a jumble of random words. The grafs had sentence variation, color, plot development.
He did another query: Think of a title for a children’s holiday book aimed at a Black audience. Within seconds, a list populated — and the winner was something to the effect of “Zahra’s Big Holiday Surprise.” I was stunned.
Advertisement
My friend smiled. “The future is already here,” he said. “You might as well get ahead of it as a writer.”
Follow this authorKaren Attiah's opinionsI thought of that convo as the news came down this week that the popular electronics site CNET has been using AI to write full articles. Frank Landymore at the Byte documents how eagle-eyed marketer Gael Breton figured out that CNET had quietly published more than 70 articles using AI since November, under the author name “CNET Money Staff.” Clicking on the author’s note reveals the truth:
“This article was assisted by an AI engine and reviewed, fact-checked and edited by our editorial staff.”
On top of that, it appears these AI-generated articles have benefited from Google traffic, despite the fact that Google has said it will prioritize human-generated content. Landymore wrote that a Google spokesman had clarified, “Our ranking team focuses on the usefulness of content, rather than how the content is produced.”
Advertisement
Indeed, the future is here. So what does this mean for writers and editors communicating with a digital audience? Should we be afraid of losing our jobs? Or is AI a tool that will save time for writers so they don’t have to waste brainpower feeding the internet content mill? (As Dickens once wrote, “No one is useless in this world who lightens the burdens of another.”)
CNET is not the first news-and-information organization to publish articles using this technology. The Associated Press has been using AI since 2014 to generate its articles about corporate earnings, under the justification that it frees reporters to do “higher-impact journalism.”
I told my friends that AI felt like an intellectual steroid. Writers can spend years reading the works of other writers and, over time, integrate those writers’ styles into their own work. But a bot can do it in under a minute. What does this mean for book authors? Will using AI come to be seen as “cheating”? Will writers start proclaiming they are “natural” writers, with no AI use in their work, akin to bodybuilders who choose not to use performance-enhancing drugs?
Advertisement
I’m not as widely known or read as Dickens (yet, at least! #ManifestingItIntoExistence), but does this mean that at some point, someone could program a bot to write exactly like me? As with music, writing in the English language follows certain rules to make it pleasing and memorable. If people can engineer pop music to make it as appealing as possible to the wide world, why wouldn’t it be the same for writing?
I know it’s a scary thing to think about, especially as colleagues in my field are facing layoffs and belt-tightening measures. But my friend is right: Journalists and regulators need to get with the program and think seriously about how AI can help ... or hurt.
Global Radar: The model of the future
There has long been talk about the ways tech reinforces inequalities and racial divisions. Stories abound about AI bots learning our most racist and sexist behaviors, and perpetuating them in their output.
Advertisement
A story from late last year illustrates how humans — in this case, White men specifically — can use AI to contribute to the erasure of already marginalized groups.
In 2018, the world’s first AI-generated supermodel was created. Her (air quotes) name (air quotes) is Shudu Gram, and she was created by a White man, Cameron-James Wilson. Apparently he said he used real-life supermodels Naomi Campbell, Alek Wek and Iman as inspirations. And indeed, she is dark-skinned, and her backstory is that she is a young South African woman.
Wilson has taken heat for making money from Shudu’s image, which has been used in Vogue, Women’s Wear Daily and other publications. Over at the Outlet, Hannah Warren has asked some of the right questions — I’d love to know what readers think of this:
Take the example of Shudu: She’s a Black South African woman but her creator, Cameron-James Wilson, is a white man. Does he have a right to speak on issues that Shudu does? (In his defence, he works with a woman of colour to write the content for his model.) Is Shudu now just taking jobs that BAME models have worked so hard to be able to take — and be paid for? Is she, and others like her, a way for brands to address causes of the moment without actually investing in the causes? Are models like Shudu just another way to avoid paying models of colour?
— Hannah Warren, "Its 2021 and AI Models are Officially A Thing"Home Front: The human side
I guess that in a way, the theme of this week is the tyranny of progress and perfection. Along those lines, look for a column from me next week about steroids, sex abuse and the particular dangers for women in a sport that is all about the pursuit of perfection.
Do you have questions, comments, tips, recipes, poems, praise or critiques for me? Submit them here. I do read every submission and might include yours in a future version of the newsletter.
ncG1vNJzZmivp6x7uK3SoaCnn6Sku7G70q1lnKedZLyxtc2ipqerX2d9c3%2BOaWhoaWNkrqp51qugrZ2iqHqxsdGfpqulkaOwpnnEp5%2BappOeu6h50q2cq6eZmXw%3D