For the past six months, I’ve occasionally used ChatGPT to help with some freelance work. Mostly, I treat it as a savvy thesaurus, asking things like, “What are some terms that mean ‘naive’ but aren’t condescending?” or, “How else can I say ‘innovation’ or ‘breakthrough’ that don’t overhype a medical advance or rely on cliche?”
The platform has also helped me keep my facts straight. For instance, when I needed to be precise about which GLP-1 medications are approved for diabetes versus obesity and which brand names (e.g. Ozempic, Wegovy) were associated with which active ingredients (e.g., semaglutide, tirzepatide), ChatGPT almost instantly provided a handy chart.
However, I’ve largely resisted considering how else to use AI as a freelance journalist, since I value the creative and learning processes my brain undergoes while reporting and writing. Plus, ChatGPT has shown me I am better than it at coming up with clever turns of phrase: When I challenged it to brainstorm some puns about running and Taco Bell (for a story about a fast-food-fueled ultramarathon), it churned up cringe-worthy responses like, “Legs say no, but the burrito says go.”
But during October’s Lunch and Learn, my fellow freelancers shared other ways they’re using AI as a “very smart, very fast assistant” without crossing ethical boundaries or ignoring valid concerns about privacy. Here are the key takeaways:
Get on board — or get left behind
One AHCJ member predicted that, like the dermatologists he heard from at a recent conference, AI won’t replace us, but the journalists (or doctors) who use it are going to replace those who don’t. “AI is a great tool for us,” he said. “I think of it like having an assistant for grunt work.”
Freelancers, who can make more money by working more efficiently, can particularly benefit from such an assistant. One participant called it an “absolute life-saver.”
Tap AI for time-consuming, non-creative tasks
Among the freelancers who do use AI regularly, they find it most helpful for tedious tasks that take a lot of time but not a lot of brainpower. Think looking up source email addresses (AI may have access to databases you don’t), drafting annotations, removing filler words from transcripts and reviewing contracts.
One book author said she used a paid version of Claude to help transfer annotations into endnotes, and create bibliographies for each chapter. She still had to double-check it all, but the “assistant” condensed what would have taken several days into four hours.
“There are a lot of tasks you can use AI for that aren’t unethical or morally questionable,” like asking it to create a story for you would be, one participant said. In a lot of these instances, it’s not that different than using Google — just much better.
Consider AI for interpreting dense medical studies and brainstorming interview questions
Some journalists said they like to use AI for help with breaking down jargony medical studies, thinking through interview questions, and brainstorming story ideas or potential headlines. In these cases, they consider the tools like a colleague who can advance their ideas and skills more rapidly.
Not everyone, though, is comfortable frequently turning to AI in cases like these. I, for one, feel uneasy about offloading the rewarding challenge of deciphering a study or coming up with a fresh story idea. If I enlist AI, will those mental muscles weaken? It’s a question each journalist can consider for themselves, and revisit often. What feels right may depend on the assignment and your schedule.
Be aware of privacy concerns
Uploading an entire draft of an original story to ChatGPT for fact-checking or copy-editing, for one, isn’t the smartest in large part because the platform lacks privacy protections. “Think about: ‘Could this be subpoenaed?’” one AHCJ member said. That’s especially important if you’ve written a sensitive story with anonymous sources.
One potential workaround is upgrading from a free platform to a paid tier on something like Claude, several freelancers said. Do your research on what is and isn’t private before feeding it delicate information.
Follow best practices
Keep in mind that AI can only pull from material that’s already out in the world, so if, for example, you ask it to come up with some headlines, there’s a chance it could be plagiarizing an already published story. Similarly, if you ask it to write in your voice, it could be plagiarizing you.
If you use AI to identify sources and relevant studies — which several attendees said they do — be sure to also use your own judgment. It’s important to make sure you’re diversifying sources and not merely quoting people and research that past stories have, too.
Another blatant concern: AI can “hallucinate” or make up facts, people and studies that don’t exist. So, it’s critical to corroborate everything it says.
For the best results, members said they explicitly ask AI not to make anything up. They tell the tools what type of job they have, and what their goals are for the prompt (“I’m a journalist writing for a physician audience. Is there a way to make this sentence more concise without dumbing it down?”) And, they largely turn to it for small, not big, tasks — for now. “Like assistants,” one participant said, “they need very clear instructions.”
Source: 5 tips for maximizing AI as a freelance journalist
