When filmmaker and creative director David Mesfin typed "Black man with a surfboard" into an AI image generator, the results were consistently disappointing. Instead of generating authentic images of Black surfers, the artificial intelligence program would produce images of white men with darkened skin holding surfboards. This glaring technological blind spot revealed a fundamental problem: Black surfers simply weren't part of the AI's training data set.
The discovery was particularly striking for Mesfin, who had just finished producing "Wade in the Water," a documentary exploring the largely overlooked 1,000-year history of Black surfing. The irony was acute - even after creating a film that documented extensive Black participation in water sports, cutting-edge AI technology had no evidence of their existence. As Mesfin investigated further, he found the absence of Black and other BIPOC (Black, Indigenous, and People of Color) faces extended far beyond surfing imagery to encompass outdoor activities, workplace settings, schools, and homes.
Rather than accepting this technological limitation, Mesfin saw an opportunity to fundamentally rewire how AI perceives diversity. Partnering with Pocstock, a comprehensive BIPOC imagery database, and collaborating with dozens of creative agencies, Mesfin and his team at Innocean launched "Refacing the Future." This groundbreaking initiative aims to retrain popular AI models used by industry giants like OpenAI, Google, and Microsoft to better recognize and generate images of people of color.
"We need a broader group of people that can actually look at this technology and what is being developed," Mesfin explained. "If you're going to create a tool for the world, it should represent the world accurately, right?" The project represents the first comprehensive effort to create an AI guidebook specifically focused on diversifying datasets for years to come.
The outdoor recreation industry has long struggled with representation issues, with generations of BIPOC skiers, surfers, hikers, and climbers watching white athletes dominate videos, advertising campaigns, and magazine coverage. Mesfin experienced this firsthand growing up around St. Augustine, Florida, where he consumed media featuring athletes who didn't look like him during his own surfing experiences. Eventually, he discovered other surfers of color and realized that mainstream media simply reflected the perspectives of its predominantly white creators.
Mesfin believes AI suffers from similar limitations rooted in its creators and training data. "Speaking to under-representation, I think it's simply because of the individuals that are behind creating the content or technology," he noted. Throughout generative AI's brief history, those individuals have typically been both white and male. Additionally, these machine learning models draw from previously published material that often perpetuates existing biases and stereotypes across generations.
"Media has always painted a negative image of BIPOC communities - we've never been represented in our true sense," Mesfin observed. "[AI] is picking up on those things, those stereotypes - it's reflecting what's already out there." When algorithms fail to recognize outdated depictions of people of color, they frequently misidentify them entirely. A 2019 federal study concluded that Asian and Black people were up to 100 times more likely to be incorrectly identified or depicted by AI models compared to white men.
Despite technological improvements in recent years, significant gaps remain. Innocean points to a striking example: Asian dentists comprise 22 percent of the entire U.S. dentist population, yet when AI generates images of dentists, Asian people appear only 2.3 percent of the time. Such disparities highlight the urgent need for systematic correction of AI training data.
Refacing the Future addresses these shortcomings by strategically flooding AI systems with diverse, properly categorized information. The alliance with Pocstock provides access to over 1.8 million images of people of color, but Mesfin considers the database's organizational system the key to creating a more equitable digital future. Each image in Pocstock's collection is manually tagged for skin tone, gender, race, age, and various cultural identifiers, making it easier for AI models to identify and incorporate these details into automated image generation processes.
The initiative has identified 22 creative agencies to produce BIPOC-specific media content, not only hosting imagery and video on the Pocstock platform but also tagging and making that work available to OpenAI, Canva, Google, and other AI models hungry for diversity data. According to the Refacing the Future website, over 16 BIPOC photographers are working to produce 96,000 pieces of original media within a year, featuring Hispanic skateboarders, Native American water sports athletes, Asian snowboarders, and many other historically underrepresented groups.
As the database continues expanding, Refacing the Future has prioritized establishing guidelines for sustainable growth. The project's guidebook encourages creative agencies to organize photoshoots with diverse talent, including photographers and other crew members behind the cameras. Additionally, the guidebook provides creative resources and cultural curators to help media producers navigate more inclusive approaches to content creation.
Currently, AI operates as a powerful tool without clear ethical guidelines, which Mesfin and his collaborators view as an opportunity to establish best practices. "I want to hand agencies this guidebook and say, 'Hey, here's how we did it,'" Mesfin explained. "Here's how you can find photographers, how you should approach your photoshoots. That way it can be more a reflection of the community and authentically capture the demographic."
While still in early stages, Refacing the Future has identified several additional areas needing improvement, including representation of women in STEM fields and evolving roles of women in household settings. When Mesfin was selected to produce the platform's inaugural creative shoot, he returned to familiar territory: the ocean.
On a sunny February morning, Mesfin and a team of photographers and surfers of color gathered at Huntington Pier in Southern California. Despite the morning chill from ocean spray, warm air surrounded the crew as they approached the iconic wooden pier structure. The moment felt like completing a circle for Mesfin, who was now actively filling the representation gap he had identified during his frustrating internet searches years earlier.
Photographer Kory Lamberts, who met Mesfin at "Great Day in the Stoke" - the world's largest gathering of Black surfers - felt inspired by the creative director's vision. When invited to participate in this first Refacing the Future project, Lamberts knew he had to be involved. "Here was this chance to reshape and recreate the future from the lens of people that haven't been able to tell it yet," he explained. "It's time we figure out a way to move forward with these technologies to utilize them for our communities."
With four athletes in the water and photographers positioned both on land and in the surf, the team captured extensive footage and still images that would later be uploaded to Pocstock and integrated into today's popular AI modeling programs. Through this work, they are ensuring that digital technology finally recognizes exactly who they are and their rightful place in outdoor recreation and beyond.
