AI won't steal your writing job (but it will change it). Here's how – USA TODAY

“Shakespeare’s not such a great writer,” a fellow student said during an English class years ago. “His stuff is lazy! There’s one cliché after another.”
Certainly, the then-teen could be forgiven for thinking that the playwright William Shakespeare phoned it in, so to speak. His plays are peppered with phrases that are now clichés. “My own flesh and blood,” “cruel to be kind” and “method to my madness” are a few from “Hamlet” alone.
But those tidbits weren’t clichés before Shakespeare. They didn’t exist until everyone saw that his phrasing was so imaginative and poignant they couldn’t resist adopting them.
The distinction between turning a phrase and borrowing one is critical to gauging where generative AI is heading, and what threats and opportunities it may present for the future of human composition.
Schumer AI crack down:Senate leader Schumer unveils plans to crack down on AI
Large-language models – generative AI platforms like ChatGPT from OpenAI and Google’s Bard – are learning to produce compositions based on very specific, complex requests.
They draw from published work, and leverage that to assemble Shakespearean-like manuscripts, term papers or virtually any type of document – in just seconds.
For all its early well-documented blunders, generative AI is an undeniably disruptive, transformational technology that will evolve to change a host of white-collar professions, from legal and medical to academic and even travel. And writing.
More:Google launches ‘experiment updates’ page for Bard, says chatbot math skills are improving
AI too ‘woke?’:Why Elon Musk wants to build ChatGPT competitor: AI chatbots are too ‘woke’
AI’s limitations will keep humans at the top of the food-for-thought chain. For one – and with apologies to Shakespeare – a human both a lender and borrower be. Generative AI, on the other hand, can only borrow – from us.
For all the copy it can generate, it flat-out can’t invent fresh, new expressions and observations. It needs us for that.
Writers have prodded some eyepopping gaffs out of ChatGPT, the Kleenex of generative AI platforms. AI learns about connections and associations much differently than we do. Something so simple for humans to grasp, like my Kleenex reference, can be very difficult for AI to parse.
The more mistakes we identify, the more the technology will improve. But the possibility of misreads that convince generative AI to drive confidently off a cliff will never disappear entirely. So we’ll never hand over the keys.
Speaking of keys, AI has a track record of falling short of full autonomy across many applications. Like self-driving vehicles.
In 2016, many big-name automakers predicted that autonomous vehicles would be everywhere by 2021. Some manufacturers were even considering interior designs without steering wheels.
Obviously, that didn’t happen – and probably won’t. Tesla aside, many carmakers since have scaled back or scrapped development.
But AI is taking on specific tasks that help us drive better. It can warn of potential hazards and brake if we don’t respond quickly enough. It can even take the wheel to parallel park or cruise along boring, well-characterized stretches of highway.
Radiology is taking a similar approach. AI is very good at spotting some abnormalities on X-rays and MRIs. At the same time, there will always be new combination of angles and shades on an image that can trip up the technology.
So while radiologists increasingly are leveraging AI for the tasks it does well, they are not ceding control. The radiologists are in charge.
That’s the model I see developing for writers. Bing AI, Microsoft’s ChatGPT-based offering, and Bard are great first stops for any assignment. They can help point out the relevant issues and suggest a structure.
I should have taken my own advice. Just now, I asked both ChatGPT and Bard to write an essay about the future of writing in the face of generative AI, and they both performed well. They called out many of the issues I’ve raised here.
But I’ve been writing on and off for days. And the AI took just 10 seconds. Pretty jarring.
Clearly, generative AI is both an opportunity and a threat. Which one, really, is what we choose to make of it.
In the words of Shakespeare’s Hamlet: “There is nothing either good or bad. But thinking makes it so.”
USA TODAY columnist Mike Feibus is president and principal analyst of FeibusTech, a Scottsdale, Arizona, market research and consulting firm. Reach him at Follow him on Twitter @MikeFeibus.


Leave a Comment