AI image generators are one of the hottest new developments in tech, and every social media platform — including Facebook — has raced to create their own. However, these tools are only as good as the data they’re trained on, and according to experts, that can include potentially offensive information from around the web. This week, a reporter from The Verge exposed that in embarrassing detail by proving that Facebook’s AI image generator tool couldn’t generate an image of an Asian man with a white woman.

Much like how human biases affect our everyday interactions, experts say those same biases can leak into artificial intelligence, but at a massive scale.

“The image generator not being able to conceive of Asian people standing next to white people is egregious,” Verge reporter Mia Sato wrote in her piece. “Once again, generative AI, rather than allowing the imagination to take flight, imprisons it within a formalization of society’s dumber impulses.”

For its part, Facebook knows that its tool is a work in progress, including a disclaimer on the site that says images it generates “may be inaccurate or inappropriate,” also noting in a blog post that “potential bias in generative AI systems is a new area of research.” However, it might not have been a great idea to let a tool capable of this kind of error out into the wild so soon.

Recommended Resources

Choose what the experts use: award-winning cybersecurity you can trust and rely on.

Surf the web truly incognito. Try Bitdefender Premium VPN, the ultra-fast VPN that keeps your online identity and activities safe from hackers, ISPs and snoops.

System Mechanic 14 – Make your computer run like new. Winner of 200+ Editor’s Choice awards!