I had an interesting situation last Friday that reminded me that when I say, “AI is always changing,” it really is true. AI is always changing.

About six months ago, I pitched a webinar to several genealogical societies about creating Custom GPTs for genealogy in ChatGPT. Custom GPTs are customized versions of ChatGPT that you can set up with specific instructions, reference materials, and capabilities so they can perform the same kinds of tasks over and over again. They can also be shared with other users.

I was scheduled to present the course for the first time last Saturday, and again to a second society a few days later. I was excited to see how it would be received. If the webinar was successful, I planned to add it to my available courses for 2027.

So, last Friday, I opened the presentation to review my slides, prepare my live demonstrations, and double-check which ChatGPT subscription levels could create Custom GPTs. I immediately ran into a roadblock.

The functionality I planned to demonstrate was no longer available to most individual ChatGPT users. OpenAI had changed Custom GPT creation so that new GPTs could no longer be created with Free, Go, Plus, or Pro accounts.

Here I was, about 18 hours before my first webinar, preparing to teach something that neither I nor most of my audience could actually create!

A conversation with ChatGPT led me to Projects, a feature that could accomplish many of the things I wanted my genealogy audience to learn. The webinar was quickly reworked around Projects, and the crisis was averted.

But that wasn’t the end of the story. During that conversation, ChatGPT also mentioned something called Work, which I had never used but immediately wanted to investigate. In my initial search for GPTs, I had also come across Plugins, another feature I hadn’t explored. Then, while relating the story to a member of my local genealogy group, he mentioned Skills. That was new to me, too.

Here I was, the tech expert who uses ChatGPT almost every day, suddenly learning about three features I had either ignored, glossed over, or simply never encountered.

It reminded me of driving to work every day and always taking the same route, never realizing that an entire new complex had opened two blocks away because my usual route never took me past it.

That’s one of the challenges of keeping up with AI. We tend to learn the features we need, develop our favorite ways of working, and then keep traveling the same familiar roads. Meanwhile, developers are building whole new neighborhoods just out of sight.

And keeping up can be a lot, especially if you use more than one AI program. In my day-to-day work, I primarily use ChatGPT with a Plus subscription, along with occasional use of Claude on a free account. I also follow several groups, newsletters, and publications that help me keep tabs on what’s changing.

Genealogy and Artificial Intelligence: This Facebook group is private, although “private” is a relative term when there are roughly 27,000 members! Discussions span the length and breadth of AI use in genealogy. You’ll find beginners asking for advice about their first prompts alongside advanced users demonstrating sophisticated AI workflows and agents. The group is friendly and supportive of different skill levels, and familiar names in genealogy and AI, including Blaine Bettinger, Steve Little, and Mark Thompson, are active participants.

Substack: Contrary to how it may seem when you’re first creating an account, Substack has plenty of content you can read for free. One newsletter I recommend is Ethan Mollick’s One Useful Thing. Mollick is a professor at the Wharton School who has been researching, experimenting with, and writing about AI for years. His newsletter makes complicated developments understandable and frequently includes entertaining examples. His book, Co-Intelligence: Living and Working with AI, is also an excellent introduction to thinking about how humans and AI can work together.

For a genealogy-specific perspective, Steve Little’s Vibe Genealogy focuses on AI tools and techniques specifically for family historians. It’s a useful way to see how developments in the larger AI world can translate into actual genealogy workflows.

Traditional Media: If you already subscribe to publications such as The New York Times or The Washington Post, their technology coverage is worth following. Both cover much more than AI, of course, but that broader perspective can be useful. I’m also a big fan of MIT Technology Review, which covers a wide range of emerging technology and helps put individual AI developments into a larger context.

The lesson I took from all of this is that missing tech news, especially AI news, is normal. The information is coming at us fast and furious, and we shouldn’t expect ourselves to keep up with every new feature, model, tool, and announcement.

Sometimes the most useful thing we can do is simply take a different road once in a while and see what has been built two blocks over.

And I’m very glad I took that detour before I found myself explaining to a live audience why the Custom GPT button in my demonstration had disappeared!