Why I’m Still Not Sick of ChatGPT

(But maybe you are)

I pasted this article into ChatGPT and just asked it to make an illustration using DALL-E. This is what it made.

The other day I saw a post on Mastodon by Harvard Library curator John Overholt that said:

“I asked ChatGPT” is the new “Webster’s Dictionary defines” of crappy introductory sentences.

It’s so true. Writing about ChatGPT is becoming lazy and cliché. And we got there very quickly. ChatGPT was only out for a few weeks before we got sick of news stories that included variations on “Believe it or not, that introduction was actually written by a computer.” So many articles and stories on the topic quickly all sounded the same.

And it all looks the same, too. If I see a post with a screenshot of a ChatGPT conversation, my eyes glaze over and I struggle to pay attention. You might as well be reading me a poem or telling me about the dream you had last night.

And yet, I still find new ways to use ChatGPT every day that make my life easier. It continues to amaze me. And I think, “I have to tell people about this new way I’m using it!” But then I realize nobody wants to hear that anymore. Or do they? Have normal people shrugged and moved on, or is everyone still tinkering with it like I am? Maybe the use case I stumbled on has been stumbled on by everyone else already, and we’re all just sick of talking about it.

But it truly has been helpful to me in ways that might be helpful to you, or get you thinking about useful ways to use it when the novelty of generating stories in the style of famous people wears off.

So, at the risk of making your eyes glaze over:

I asked ChatGPT…

Make Order From Chaos

My neighborhood has a pretty active Listserv, and somehow it remains mostly drama-free. One of the things I use it for most frequently is getting recommendations. I recently needed to find a new vet to get a second opinion for a sick cat. So I asked the Listserv for help. And I got around twenty responses.

That’s twenty emails, each written differently with no consistent format, telling me about veterinarians nearby. Some were duplicate suggestions. Others had stories that went along with them. It was a bit overwhelming. I didn’t know where to start.

So I asked ChatGPT. I pasted in the entire email thread and said:

There are responses to a question I asked that came via email. Please go through them and create a table with the following columns: Business/person who was recommended; Contact info and location if provided; who recommended them; any additional details or notes. If multiple people made the same recommendation, only list that business once. List them in order of who was recommended the most.

It gave me a table that looks like this:

My neighbors’ last names have been removed for this newsletter.

Wow. That is so much easier for me to digest! And I can copy-and-paste that whole thing into my Notes app as an editable table for future reference.

When other people from the Listserv reach out and say, “Can you share with me the responses you get?” I just email them the table and they’re always impressed by how neat and organized I am.

I have since repeated this with other recommendation requests. Even when I don’t get so many responses, it helps me save the information neatly in one place for later. Here’s another example from when I needed a plumber recommendation:

That’s so much better than sifting through a mass of emails!

Putting Doctors on a Map

Our pediatrician recently suggested that one of my kids see a specialist (he’s fine, don’t worry). Since specialists are often booked far in advance, he emailed me a huge list to go through of people he recommends. There were 49 names on the list, with phone numbers. No addresses or anything else. He said they’re all good.

What am I supposed to do with 49 names and phone numbers? Start at the top and work my way down? Some doctors are more than 30 minutes away, and some are 10 minutes away. All else being equal, I’d like to visit the closest doctor. I was a bit paralyzed by so many choices with no idea where to start.

So I asked ChatGPT for help.

First it converted the list to a simple spreadsheet. Then it walked me through the process of getting a Google Places API key. Finally, it wrote a Python script that uses the spreadsheet and the API key to look up the doctors based on their names and phone numbers. I ran the script and it spit out a file I could import to Google Maps with all the doctors’ locations plotted.

Clicking any of those pins gives me a popup with the doctor’s name and phone number. This gave me a much better starting point for finding a doctor not too far away. By starting close to home and working my way out, I was able to easily find a doctor we could fit in our busy schedules.

Those are a couple examples of why I continue to be excited about A.I. for things that don’t have to do with making funny images or chatting with waifu girls or whatever. It actually is useful in my life, saving me time and stress.

But there’s still a bit of friction. I had to get an API key, generate a script, and either know how to use such a thing or be comfortable following its instructions walking me through it. In a few more years, when it requires less hand-holding in both directions and it can just say “Give me the list you have and I’ll plot those doctors on a map for you. I can also check the doctor’s ratings, whether they’re accepting new patients, and if they take your insurance,” it will be even better.

And speaking of things that are even better: If all goes according to plan, the next issue of this newsletter is going to be something big that I’ve been working on for months (and has nothing at all to do with ChatGPT). I’m a bit nervous saying this out loud because I could easily get delayed, but hopefully saying this publicly will help get it done on schedule. So stay tuned.

As always, thanks for reading. See you next time!

David