Register for PRSA’s new Intersections Conference, which features three distinct Learning Tracks with programming specifically tailored for the needs of technology, financial, entertainment and sports communicators. Attendees will select their Learning Track during the registration process, and will join a series of breakout sessions curated by leaders in each industry. For more information: http://bit.ly/
Diane S. Thieke
Robots are coming. But are they coming after writers? Until now, the scenario hasn’t seemed likely. After all, experts predict the tasks most likely to be automated are those that are predictable, routine and repetitive.
Writing and editing, of course, require a great deal of creativity. That’s why your tax preparer has a 98.7 percent chance of seeing his or her job automated in the next 20 years, while PR specialists and editors — at 17.5 percent each — are less likely to be replaced, according to a 2015 NPR survey. Writers are safest of all, with a 3.8 percent chance of being supplanted by automation.
Still, the uncomfortable truth is that algorithms can handle some writing and editing tasks, such as correcting misspelled words. And fluid though grammar can be, it still consists of rules, which can be turned into computer code.
Indeed, it’s possible that robots might come for our jobs someday. More likely, though, they’ll transform the writing process instead. Since AI and machine learning could help us do our jobs better and faster, why not take advantage of these technologies and automate the more routine aspects of writing and editing?
Revising, clarifying and proofreading
Let’s be honest: The writing process is labor-intensive and impossible to scale. It requires an incredible amount of creativity, thought and energy. Sure, it’s possible to write a thousand words in under an hour. But the resulting quality will not meet your standards or those of your clients.
Until recently, I didn’t see how automation could be useful in my writing life. I resisted the technology because early checkers for grammar and spelling were poor substitutes for human editing. But over the last few months, I’ve incorporated three tools — Grammarly, Hemingway Editor app and ReadablePro — into my writing process, with some success. I’ve come to believe that AI can give us more time in the day — time that could be spent on higher-value tasks, such as cultivating client relationships. And in a surprising twist, I’ve discovered that artificial intelligence can improve content quality.
To be sure, these tools aren’t capable of doing the heavy lifting required during the rewrite stage. But they can help in three important phases of the writing process: revising the first draft; copyediting for clarity, grammar and spelling; and proofreading the final draft.
Keeping or rejecting suggestions
Like most writers, I begin every article I write by typing my thoughts as fast as I can. The first draft is always a mess. But Grammarly cleans up my typos in just a couple of minutes, producing a draft that I can work with. The free version of the software checks for common grammar and spelling errors, such as subject-verb disagreements and contextual spelling mistakes. But the premium upgrade (which requires a $139.95 annual subscription) evaluates sentence structure and offers vocabulary suggestions. Grammarly throws a lot at you at once, but the decision to keep or reject its recommendations is yours. Clicking through each change can be cumbersome, but I’d rather be in control than let the software make wholesale revisions I don’t want.
Eliminating unnecessary words
That lousy first draft usually undergoes several heavy rewrites. By the time I have near-final copy, I’m too close to it and have lost the objectivity required to recognize overly complex sentences or excessive adverbs.
The Hemingway Editor app (free on the web; $19.99 for the desktop version), inspired by Ernest Hemingway’s spare writing style, analyzes text for readability. In addition to identifying hard-to-read sentences, it also highlights adverbs, recommends simpler words and flags passive voice. When I drop my draft into the app, it shaves time from the copyediting process by marking problems in the text, so I can focus my energy on simplifying those sections. If you’re a fan of Strunk and White’s writing guide The Elements of Style, you’ll like this app.
Detecting plagiarism and clichés
Although I wouldn’t use it for a final read, Grammarly reduces the time I spend proofreading by pinpointing errors I’ve overlooked or missed. But there’s another reason I turn to Grammarly at this stage: its plagiarism detector. The company claims the software checks billions of pages to identify identical copy. As I often reference the work of others to back up my arguments, this check offers some reassurance that I haven’t cited too much of another person’s writing.
Another app, ReadablePro, can improve the quality of your prose. It scores copy according to 10 readability tests, including the Flesch-Kincaid Grade Level and the Coleman-Liau Index. Each test serves a different purpose, so you’ll need to select those that are most relevant for your work. But ReadablePro helps ensure your copy is appropriate for the reading level of your target audience. I like the tool for its ability to notice clichés and buzzwords — something the other two apps don’t do.
Assessing opportunities and pitfalls
The results of my experiments with AI writing tools have been encouraging. I estimate that I’ve reduced the time I spend writing and editing by 10 percent. My copy is sharper, too.
But none of these tools is perfect, and they have plenty of pitfalls. For one thing, I disagree with roughly 30 percent of the edits they suggest. For example, ReadablePro tells me that words like “exceptional” and “relationships” are difficult or too long. Whether I keep such words depends on my target audience.
More troubling is that sometimes the algorithm will recommend a change that alters the meaning of the sentence. Without human oversight, there’s a risk that robots will water down messages or get them completely wrong.
As writers, another risk we face if we rely on current AI tools is that all of our work will start to sound alike in style and tone. Thought-provoking, beautifully crafted copy should be lyrical and distinct, and inspire readers to act. At the moment, only a human can produce this type of content. We’re still a few years away from when writing and editing robots can create remarkable work.
For now, though, writers and editors should feel comfortable embracing AI tools. Any method that helps streamline the editorial process will benefit our businesses and clients.
Diane S. Thieke, APR, is a freelance writer and editor who works with global B2B companies in technology and media. Previously, she led editorial, PR and marketing teams at Dow Jones. In 2009, she won a PRSA-NJ Pyramid award for her eBook on PR measurement.
This article originally appeared in the June 2018 issue of Strategies & Tactics.