Wonder what your dog would look like as a cat? There’s a new AI tool for you.

If you’ve ever pondered what your dog might look like as a cat, or another dog, your time is now. And even if you haven’t, the tech company NVIDIA wants to show you.

NVIDIA has leveraged big gains in AI technology to develop a fun tool called the GANimal App, which allows users to upload a photo of their pet and generate pictures of what it would  look like as either different breeds of the same animal — or a different animal altogether. 

After watching the demo, I decided to give the tool a whirl with a photo of my dog, Charlie. 

Here’s the original.

The results were, well, varied. 

Some of these, like the flat-coated retriever (No. 3) and the Irish wolfhound (No. 8), actually resemble Charlie. Even the meerkat (No. 13) bears a strong resemblance. 

But for the most part, the dogs just look like dogs, and some of the other animals look pretty much like themselves, with Charlie’s tongue added. The tiger cat (No. 16) looks like any old cat, look of contempt and all. 

You can keep hitting the translate button to generate more results, but the resemblances remain about the same. 

Some of those, especially the corgi (No. 16) and the Leonberger (No. 12), look like Charlie. The Persian cat (No. 11) looks like a cat. 

Still, it’s a fun exercise — and an interesting way for NVIDIA to flex its AI developments. Thanks to developments in the Generative Adversarial Network, which NVIDIA describes as “an emerging AI technique that pits one neural network against another,” we have “new AI techniques that give computers enough smarts to see a picture of one animal and recreate its expression and pose on the face of any other creature.”

The research team that developed this algorithm explains in a paper how they were able to create a system that generates a new animal image using a single photo of the original animal: 

“In this case, we train a network to jointly solve many translation tasks where each task is about translating a random source animal to a random target animal by leveraging a few example images of the target animal,” Liu explained. “Through practicing solving different translation tasks, eventually the network learns to generalize to translate known animals to previously unseen animals.”

It’s only a matter of time, then, until we’re faced with a burgeoning crisis of animal deep fakes. But, for now, enjoy the fun.