Dog rescue Wags & Walks wants to know if you can spot the real puppy in a pack of artificial intelligence (AI) generated dogs. The rescue with locations in Nashville and Los Angeles features the game This Dog Exists. This Dog Exists is designed to promote the adoption of homeless pooches. The game is a clever use of artificial intelligence that serves the greater good.
This Dog Exists was created by Matt Reed, the creative technologist at the ad agency Redpepper. The page aims to reduce the population in dog shelters. Mr. Reed told Fast Company,
Every year we participate in an advertising industry event called Createathon where we donate work for nonprofits over the course of 24 hours … We stay up all night, and it is a blast. We also get very delirious, which is partly where this idea came from.
Mr. Reed says they also wanted to try and do something to bring some extra awareness to their cause.
We explored a few different ideas but kept coming back to doing something with [the open-source AI generative image algorithm] Stable Diffusion … Contrasting AI dogs versus real dogs seemed interesting.
This Dog Exists is designed to be simple. It displays a grid of four dogs; one of them is real, and the other three were created by the Stable Diffusion AI. FC reports that the team used Lexica.art—a generative image search engine—to find the perfect prompt for the cutest puppy creation. If you fail to guess the real dog 10 times, This Dog Exists displays a message that says,
Woof! It looks like you need a real dog because you don’t know what a real dog looks like 🙂
Good for dogs
While Stable Diffusion, DALL-E, and MidJourney still have problems creating believable human faces, Stable Diffusion seems to be pretty good at conjuring dogs. Except for a few imperfections, the fake dogs look pretty much like the real thing.
This Dog Exists isn’t groundbreaking in its use of AI technology, but it is a cute and fun way to engage people. “Even if it only gets one extra dog adopted, I’ll consider it a success,” Mr. Reed concludes.
AI programs called generative adversarial networks, or GANs. GANs were designed by researcher Ian Goodfellow and his colleagues in the year 2014. They can learn to create fake images that are less and less distinguishable from real images, by pitting two neural networks against each other. Researchers says that artificial intelligence can create such realistic human faces that people can’t distinguish them from real faces – and they actually trust the fake faces more.
rb-
Needless to say, the ability to generate realistic faces raises all kinds of ethical questions, even if they don’t belong to real humans.
These massive-scale machine learning systems can harm marginalized people through deeply embedded biases that can’t be easily engineered out. AI’s frequently create racist and sexist stereotypes.
While AI engineers say they’re doing their best to create safeguards that prevent abuse, it’s likely we’ve only just begun to see these large AI systems are capable of—and what types of harm they might cause.
Related article
Ralph Bach has been in IT long enough to know better and has blogged from his Bach Seat about IT, careers, and anything else that catches his attention since 2005. You can follow him on LinkedIn, Facebook, and Twitter. Email the Bach Seat here.