Hollywood Won't Cast Asian-American Stars, but A.I. Machine Learning Can

In the movies, Captain America is a white guy. But what if he were played by Korean-American actor John Cho? This was tricky to imagine a few months ago, as movie studios remain allergic to casting Asian actors in starring roles. Now, thanks to a new collection of “deepfakes,” seeing is believing as a new project illustrates what Asian-Americans might actually look like as the heroes of the biggest Hollywood blockbusters.

Digital strategist William Yu is the creator of #SeeAsAmStar (short for “See Asian-American Stars”), a collection of videos and GIFs published on May 4. The title is the goal, as Yu tells Inverse he wants to “break that notion and assert that we can play anything.”

“We are human,” Yu says, “and have all the qualities of any other person that can convey emotion that run the gamut of the human experience.”

#SeeAsAmStar is a spiritual sequel to Yu’s last project, 2016’s #StarringJohnCho, which used photoshop to place actor John Cho (Star Trek) on the posters of romantic comedies and James Bond movies. #SeeAsAmStar expands Yu’s technical scope and subjects, featuring Constance Wu (Fresh Off the Boat), Arden Cho (Teen Wolf), and Steven Yuen (The Walking Dead) swapping for Scarlett Johansson, Jennifer Lawrence, and Joseph Gordon-Levitt in their most famous roles. In this project, Yu uses John Cho again, this time swapping him for Chris Evans in the Marvel franchise. (None of the actors have any affiliation with Yu.)

The goal is to show Asian actors in the archetypes usually reserved for white actors, from the superheroic to the hopelessly romantic.

Yu isn’t an obsessed fan trying to get his favorite actors more work. Even when he made #StarringJohnCho, he says “it wasn’t really about John Cho,” but “about making a statement, about us being okay and getting used to what it means to see an Asian face in a lead role.”

Yu’s newest striking images weren’t like his previous project. He didn’t simply use photoshop this time. Instead, Yu used A.I neural networking, or Deep Learning Technology, where the term “deepfakes” originates, which uses a network of algorithms informed by a database of images. And it is scary easy to collect thousands of images of faces, through Google, YouTube, stock photos, and social media.

“With hundreds of face images, I can easily generate millions of distorted images to train the network,” told the anonymous creator of deepfakes in a December 2017 interview on Motherboard. “I feed the network someone else’s face, the network will think it’s just another distorted image and try to make it look like the training face.”

It’s a lot more sophisticated than “face swapping,” but it’s relatively simple to learn. In that same Motherboard pierce, A.I. researcher Alex Champandard said deepfakes can be done with consumer-grade computers. “This is no longer rocket science,” he said.

For #SeeAsAmStars, Yu studied dozens of movies and collected thousands of images of each actor, and sorted them through the software Fakeapp. “You have to extract the faces from each image, sort the data and clean it up [so] you’re not substituting Scarlett’s face with a hand or something,” he explains.

Unfortunately, the low barrier-to-entry for deepfakes has given it a bad reputation. In early 2018, expert users populated Reddit with deepfakes of celebrities and regular people’s faces pasted onto porn. The law has not caught up with how deepfakes infringe on privacy or defamation, but the applications of the tech remain immeasurable. Yu, at least, has a more earnest purpose for his deepfakes.

“The movies I loved didn’t star or feature people who reflected me,” he says. “This is something I’ve grown up and had a tension with. The past few years I’ve been finding ways to spark a conversation so people can realize Asian-Americans can be represented in media.”

Yu started his project in 2017, when deepfakes of Nicholas Cage in Raiders of the Lost Ark and Man of Steel went viral. “Coming from #StarringJohnCho, knowing there was a relationship inserting faces where they didn’t belong connected right off the bat for me.”

Yu has thought about Asians in media all his life. While growing up in Boston, he had an isolating upbringing as “one of three Asian families” in his neighborhood. He sought refuge in pop culture, but he was constantly reminded how much he didn’t belong. He rarely saw himself reflected on the screen, and when he did, it was usually a gag (see: Sixteen Candles). “This idea I was an outsider and didn’t fit in was always a problem,” he says.

Why aren’t more Asian-American actors cast as Hollywood leads? The problem is ouroboros, mostly systemic, and yes, racist. Studios don’t cast Asian actors under the pretense they don’t draw box office money. But Asian actors don’t draw money because they’re rarely in movies. In a 2017 L.A. Times interview, Oscar-winning filmmaker Chris Tashima put it simply: “Why aren’t there any Asian-American stars? You need to cast the person for it to happen.”

In her book Reel Inequality: Hollywood Actors and Racism, sociologist Nancy Yuen Wang was told by a casting director that Asian actors are hard to cast because “directors feel as though they’re not very expressive.”

“They’re very shut down in their emotions,” says Wang’s source. “[I]f it’s something were [sic] they really have to act and get some kind of performance out of, it’s a challenge.”

“It’s scary looking at the pool of Asian-American talent,” says Yu. “There isn’t many people.”

Viewing Yu’s work, the project feels like a cutting-edge response to white-washing, Hollywood’s stubborn tradition of casting white actors in Asian roles. One of the films used by Yu, 2017’s Ghost in the Shell, brought white-washing to the forefront when it put Scarlett Johansson in the role of an Asian woman. With #SeeAsAmStar, Yu makes a salient point with his proof of concepts, showing what a giant leap one small step makes. “There’s a bigger message beyond individual actors,” he says.

“A project like this matters to those who don’t see themselves represented. A project like this can bring it one step closer to reality. You can’t imagine it? I did it for you. Take a look.”