One of my favorite things about 3D printing has always been taking an idea and turning it into something I can actually hold. Sometimes that starts with a model I find online, sometimes I design something myself, and lately, more of those projects have started with ChatGPT.

I have been making quite a few multicolor lightboxes recently, and they are a great example of how AI has worked its way into my entire making process. It isn’t just about using AI to create an image. I am using it to help develop the idea, create the artwork, troubleshoot problems, and learn more about my printers along the way.

The process usually starts with an idea for what I want the lightbox to look like. I can describe that idea to ChatGPT and have it generate an image that gives me a starting point. From there, I continue working with it to make changes. I might ask it to simplify the design, remove the background, make certain elements larger, reduce the number of colors, or clean up details that aren’t going to translate well to a 3D print.

Once I have an image I like, I bring it into MakerLab and use Make My Sign to turn it into a lightbox. I can make additional adjustments there, export the model, bring it into my slicer, and send it to the printer.

Of course, getting a model onto the print bed doesn’t mean the project is finished.

I recently started printing lightboxes on a new printer and immediately ran into problems. Some of the different colored sections weren’t bonding together correctly. I made some adjustments and tried again. The outside edge improved, but I still had gaps between parts of the design. I also noticed one color bleeding into another because a little filament was staying on the nozzle as it moved across the print.

Instead of randomly changing settings and hoping something worked, I took pictures of the prints and uploaded them to ChatGPT. I explained what I was seeing, shared screenshots of my slicer settings, and used the conversation to narrow down what might be causing each problem. We worked through flow settings, overlap, walls, nozzle movement, and other slicer settings. I would make an adjustment, print another version, see what changed, and then use that new information to decide what to try next.

That troubleshooting process is actually one of my favorite ways to use AI.

ChatGPT didn’t magically fix my printer. I still needed to understand what I was looking at, decide which recommendations made sense, change the settings, run another print, and evaluate the results. What AI gave me was another resource to help me work through a problem instead of searching through dozens of forum posts or blindly changing settings.

I think that is where AI and making become a really powerful combination. A single project can start with a rough idea, move into AI image generation, continue into digital design and slicing, and eventually become a physical object. Along the way, I am constantly making decisions and solving problems. If the first version doesn’t work, I can bring that failure back into the conversation and use it to improve the next attempt.

For me, this is also a great example of what AI can look like in education. We spend a lot of time talking about whether AI makes things too easy for students or gives them answers they should be figuring out themselves. Making gives us another way to look at that conversation.

Imagine students using AI to brainstorm an idea, create an initial design, turn that design into something physical, discover that it doesn’t work exactly as expected, and then use AI to help them figure out why. They still have to make decisions, test ideas, evaluate results, and try again.

That is a much more interesting use of AI to me.

The goal isn’t to remove the thinking from the process. It is to give us another tool that helps us keep creating, experimenting, learning, and figuring out what to try next.