The other day over iced coffee with a friend, I found myself saying something I never would have three years ago: “The conversation around AI has evolved.”
When ChatGPT landed in late 2022, the topic of generative AI and writing was met with equal parts Anxiety, skepticism and hype.
But at a recent dinner event in San Francisco, I described how I created a podcast editing workflow in ChatGPT, and instead of raised eyebrows, I got nods of approval. Everyone then chimed in with their favorite AI apps, what worked best for them when writing and one person even described how they used AI to create a podcast video.
That cultural shift mirrors my own.
When I was kicking around the idea of launching a website, newsletter and podcast in 2023, a friend told me to lean hard into AI. He also advised me to openly detail how I was using it.
He was ahead of me, a crystal gazer of sorts, but I wasn’t ready to be that transparent.
Like many, I thought using AI for writing might send the wrong signal. At the time, most people were calling it a shortcut machine at best, while others found it distasteful or even dangerous.
At first, I compared AI to Grammarly, which I never liked, especially its insistence on the Oxford comma. Using Grammarly always felt like an English major was correcting me, without any sense of what I was actually writing about.
I assumed ChatGPT would give off the same vibe, and I likened using generative AI in writing to an electronic dictionary or thesaurus.
But I’ve learned that generative AI tools are more than a grammar crutch. They provide research and analysis support, help structure my podcasts and writing, generate SEO titles and slugs and have improved my headlines.
There’s that line from the movie “As Good As it Gets” when Jack Nicholson’s character says to Helen Hunt’s, “You make me want to be a better man.” Well, I’m not infatuated with AI like River Phoenix was in “Her,” but it has made me better.
I’ve also written plenty about how it’s a useful tool, and that journalists and writers would be foolish not to adopt it.
Here’s an example why: Substack recently added A/B testing for headlines and subtitles in which users can try alternative versions of the headline and send tests out to some subscribers.
After a set time, the headline with the most clicks wins.
When this A/B testing feature rolled out in August, I decided this would be a great use case for ChatGPT. I fed my story into ChatGPT with a prompt and got back a suggested alternative headline that featured an open-ended question.
I’m an old-school journalist, and an interrogative headline is the exact type I avoid. I can still hear in my head the voices of old editors I had who would say if you ask a reader a question in a headline or the opening sentence, you risk having them say they don’t know the answer either and then they won’t bother to read.
But I threw caution to the wind. I tested ChatGPT’s suggested question headline as the B option. To my surprise (and annoyance), it beat my preferred straight-news approach.
AI didn’t replace my judgment. It challenged it.
My readers are not journalists. And I know now that ChatGPT can expertly tap into how people read and click on stories today.
So I’m running all my headlines and subtitles and SEO and GEO settings through AI and I’m even, gulp, allowing that dreaded em dash into the headlines because their use has made a difference in views.
Of course, there are quirks about AI that I do not and will never agree with.
Such as AI’s preference for the em dash. Here’s a Slate article from 2011 that I continue to share to make my point whenever I debate the dash.
But at an event in July in San Jose, when the topic came up of generative AI and how users can tell if it is used in writing, I was shocked to learn I was seated next to one person who said she loves em dashes. For her, ChatGPT’s excessive use of them is great.
But they’re not for me. That 2011 Slate piece compared their overuse to a “pack-a-day cigarette habit,” and I quote that often. Although the dash lover I met in San Jose and plenty of others willingly puff on them like they’re addicted, how others write is none of my business.
Whether they use AI or grew up using the Oxford comma, it’s about taste, judgment and preference. Those are human decisions, not a machine’s.
AI hasn’t replaced my voice or my storytelling. But it has slipped into my creative process, sometimes as a silent partner, sometimes as a challenger and occasionally as an annoyance.
What matters isn’t whether AI is “good” or “bad,” but whether we’re honest about what it’s used for, what we reject and where our human judgment still leads.
Three years in, AI is no longer the bogeyman or the magic wand. It’s just another tool, one that works best when the human story is at the center.
Feature art created in collaboration with ChatGPT, another reminder that AI works best as a partner, not a replacement.
