I've seen AI enter the home in all kinds of ways, from home security cameras that recognize when a UPS truck pulls up to lighting that generates different shades when you describe your mood. But nothing is as striking as the new AI picture frames, which whip up images created by generative AI from your own inputs on color E Ink displays.

In theory, these frames could alter or create any image you choose, switch to a new look with a quick prompt or voice command and run for months on a single set of batteries.

I tested the two most advanced models from startup Fraimic ($400) and smart home company SwitchBot ($150). It would have been nice to see what they can do, but I couldn't get either of these frames working, running immediately into design issues that made them malfunction.

If you're thinking about getting an AI frame to grace your home or dorm walls, you may be rolling the dice on whether it'll work at all. Frankly, I'm not sure this particular AI foray is worth it for anyone. Let's get into the details to see why.

What do AI picture frames do, exactly?

The two frames I tried both offered different takes on generative AI -- the artificial intelligence that puts stuff together based on the commands you give it, such as the now-cancelled Sora AI from OpenAI. All they need is the right software and your Wi-Fi network to get to work (well, in principle).

SwitchBot's frame, available in sizes from 7.3 to 31.5 inches, lets you choose between selecting a picture from the gallery or uploading your own. You can then use the AI Studio (which requires a subscription) to edit the image, combine it with others, add elements like an elephant, mimic famous art styles, change colors and more.

SwitchBot reports that its frame battery can last up to two years, depending on how often you change it, and the app allows you to schedule new images far in advance so the frame can rotate looks on its own.

Fraimic, meanwhile, offers a much larger picture (18 to 36 inches). It's currently available for preorder and on Kickstarter. It not only lets you upload artwork but also give voice commands that the AI interprets to automatically create images on the E Ink canvas. (Both frames use E Ink Spectra 6.) Like SwitchBot, you can use a built-in gallery or upload your own downloaded photos.

Framic's frame has a couple of neat tricks, such as the ability to switch to a custom frame if you find one that fits and to automatically orient when switching between landscape and portrait modes, but its base features are quite similar.

Framing fails: Technical issues cut my tests short for now

I've been setting up home devices for years and can usually find workarounds when new products don't cooperate for my reviews. Here, though, I was at a loss. I can't even recommend one over the other because I couldn't get either up and running.

SwitchBot's frame had a solid start. I plugged it in and successfully connected it with the SwitchBot app and my Wi-Fi, then got as far as exploring options in the gallery. But as I was moving the frame around, I noticed it wasn't charging at all. The pulsing indicator light SwitchBot said would appear was absent.

Further attempts showed that not only was the frame refusing to charge, it also appeared unable to power up at all, even when plugged in using SwitchBot's charger or alternative chargers. Neither pairing nor rebooting worked. Without the ability to charge or connect, there was nothing more I could do with the frame.

I've tested many SwitchBot products, some of which I really like, and haven't run into problems or low-quality design materials like this before, so I felt disappointed. But there was another frame I had to try and compare to the SwitchBot experience: Fraimic's more expensive model.

Fraimic's frame was similar but a bit larger, with higher design quality -- an encouraging sign for a startup. However, other problems quickly developed.

Once charged (successfully), Fraimic's sturdier frame encouraged starting the connection process by scanning a QR code on the frame. It's a good approach, one I've used for countless smart home products, but there was a problem here: There was no QR code.

Fraimic doesn't have a standalone app to download at the moment, which threw a wrench in my usual fixes. The network guide Fraimic included for my test model had other options, such as connecting via a temporary open Wi-Fi network or manually entering a web portal for setup, but neither of those options let me continue (perhaps they were from earlier versions of the frame).

That's one of the perils of trying out startup technology: Designs and connections are often still in development and final results can be hit or miss. For me, the frame was a miss. Despite promising details, it wouldn't even let me start the setup process.

I reached out to both SwitchBot and Fraimic to see how they could help. SwitchBot acknowledged the problem but there wasn't much the company could do on its end yet, although I have submitted a support ticket. Fraimic acknowledged that firmware updates had made the QR code unusable in recent months and gave me some web portal workarounds to try and help. Unfortunately, those didn't seem to be working at the time of my testing, either, but Fraimic told me the issue shouldn't be present when the frames ship this spring.

AI picture frames aren't ready. And I'm not sure they ever will be

I'll stay in contact with these companies, and I suspect improvements to design and functional workarounds will come in time. However, I'm not sure the concept itself is worth it.

Even if your AI picture works flawlessly right out of the box, what are you supposed to do with it? Picture frame space is valuable real estate, typically reserved for pics that are dear to us or décor we're especially proud of. An E ink frame, no matter what it can generate, isn't a great substitute.

Yes, you could switch the frame to a different Van Gogh piece or art period every week, and while that's a fun novelty, I don't think it's a viable long-term role for a physical picture frame. You could upload a family portrait and modify it with new hues and backgrounds, but an app on your phone could do the same thing prior to printing the shot. You could brainstorm visual ideas for projects, but visiting a generative AI tool online is far more practical.

I'm not convinced these AI frames offer value, especially not for hundreds of dollars. Like various other AI applications, they seem to want to fill a niche that doesn't exist. If you're going to get one, buy it for fun only -- and keep in mind that results aren't guaranteed.