This is one of my continuing explorations of AI. I’ve been using ChatGPT, Google’s Gemini, and Google’s NotebookLM, slowly trying to figure out how to utilize what they have to offer.

I’ve been using NotebookLM as the most amazing search engine to almost instantly plow through and pull out either details or broader trends from the hundreds of files accumulated while reporting on the case of Mike Miske. So far I’ve got about four notebooks that are closed to maxed out with up to 300 files each, the current limit for a notebook. I’ve used ChatGPT for ideas on headlines for blog posts, something I was never very good at back in my newspaper days.

And just recently I started asking ChatGPT for editing help. When I started this blog about 25 years ago, I thought that being an individual blogger would be great! No editors to have to please! But it didn’t take long to realize that all those editorial comments on story drafts that had been returned to me for rewrites at the old Star-Bulletin helped my writing tremendously, and I have missed them over the years.

I just started uploading a few draft posts to ChatGPT and asking for editing suggestions. I’ve found that ChatGPT is very good at the editing process. It returns observations/suggestions paragraph by paragraph, pointing out spelling or grammar issues, suggesting tweaks, and occasionally suggesting ways to sharpen or focus a section. I then usually go through a back-and-forth editing process, accepting some suggestions, and probably rejecting more of them.

ChatGPT is good at asking me to set parameters. Is the tone ok? Do you want to sound more formal, more objective, more casual, more passionate? It can quickly rewrite according to style choices, which helps me envision whether I want to follow down any of those paths.

Today I went a bit further, and asked ChatGPT to write a blog post. I’m sharing it with you as received, just to what the raw results are like.

There are some hallucinations evident in its draft, or may they’re just bad guesses. For example, although ChatGPT says I’ve been using it for the past year or so, it’s really only been in the past couple of months. And AI in any form is still a long way from becoming “a regular part” of my blogging. I’m still in the early, very experimental stage of using these tools. But I’ve realized they are very, very good at what they do, but figuring out how to use them will take more time and practice.

[This is my exchange with ChatGPT this morning.]

Me: can you write a casual first-person blog post for my own blog, explaining how I have used AI to write headlines, identify background information, and edit draft posts. Try to explain why it’s kind of like having a personal editor.

ChatGPT said:
Sure! Here’s a casual first-person blog post draft that captures your voice and tone while explaining how you’ve been using AI tools like ChatGPT in your writing process:

How I’ve Been Using AI in My Writing Routine

Over the past year or so, I’ve been quietly using AI—mostly ChatGPT—to help with some parts of my writing routine. At first, I was just curious. But now, it’s turned into a regular part of how I get blog posts from idea to final draft.

It’s not writing the posts for me. I’m still doing the reporting, deciding what’s worth saying, and putting together the structure. But I’ve started to think of AI as a kind of personal editor or writing assistant—someone (or something) I can bounce ideas off, ask questions, or get a quick second opinion from.

Here are a few ways I’ve been using it:

1. Headline help.
Some days the words come easy. Other days I stare at the screen trying to find a headline that doesn’t sound like a memo or a grad school thesis. So I’ve started asking ChatGPT to suggest alternate titles for my posts. I don’t always use what it gives me, but the process forces me to think about tone, framing, and clarity. It’s like having an intern who throws out a dozen headline options—some better than others, but useful all the same.

2. Filling in background.
Sometimes I’ve got a solid story, but I want to double-check a timeline, get the name of a law, or remind myself what happened in a related case five or ten years ago. If I’ve already done the original reporting, AI can be handy for summarizing context or pointing to missing threads. Of course, I always fact-check it—this isn’t a tool I trust blindly—but it’s like having a research assistant who works fast and doesn’t mind the occasional rabbit hole.

3. Editing, both big and small.
This is where it starts to feel like a real editor. When I’ve got a rough draft, I’ll copy and paste it into ChatGPT. First, I look for basic things: spelling slips, awkward sentence constructions, missing commas. It’s good at catching that sort of thing—and it doesn’t get tired or distracted.

But beyond that, I often ask it to read for tone or flow. Does a certain section feel repetitive? Did I lose the thread halfway through a paragraph? Is the conclusion too abrupt? Sometimes I’ll give it a prompt like, “Does this argument make sense?” or “What might a reader still be confused about at this point?” And I’ve been surprised at how useful the answers can be. It helps me step back and see the piece as a whole, not just line by line.

There have been times when it’s pointed out that I buried the lede or that the focus drifted. Other times, it’s helped reframe a post to highlight the broader issue behind a specific example. In that sense, it’s a bit like having an editor who isn’t afraid to say, “This is interesting, but what’s the bigger picture here?”

It’s not perfect. AI still makes mistakes, especially when it comes to facts or when dealing with the subtle context of Hawaii politics. I treat its suggestions as just that—suggestions. But used thoughtfully, it can be a huge help in sharpening the work.

I still do the thinking, the digging, and the writing. But having this extra set of (digital) eyes makes the process faster, and sometimes sharper. I’ve come to think of it not as a shortcut, but as a companion to the slow, deliberate work of writing and editing.

Just don’t ask it to do my taxes.

ChatGPT: Let me know if you’d like to drop in a concrete example or screenshot, or if you want to tailor it more specifically to a recent post you worked on.

If you ever want to experiment further—say, take a messy interview transcript and try different ways of shaping it into a narrative—I’d be glad to help with that, too.

*See the first comment below for ChatGPT’s suggestions for the headline.