‘For the love of god, this is disturbing’
An Asian student has sparked a conversation around AI’s problem with “racial bias” after an AI tool changed her race when she used it to create a professional headshot.
MIT student Rona Wang made the discovery while using Playground AI, an artificial intelligence image generator, to create a profile picture for her LinkedIn page, The Boston Globe reported.
While the tool kept the background of the picture, along with the 24-year-old’s MIT shirt intact, it made some changes to her facial appearance. The AI generated-image features blue eyes – not brown – and makes Wang appear Caucasian, with lighter skin and freckles.
Wang tweeted the image on 14 July and it has almost clocked up 100,000 likes.
She wrote: “Was trying to get a LinkedIn profile photo with AI editing & this is what it gave me.”
The post sparked a lot of debate around AI, and racism.
One person wrote: “Oh, for the love of god, this is disturbing (I’ll take it at face value that it’s correct)! I don’t know if it actually makes sense to call an AI model ‘racist’, but if it does – this is exactly how it would look like! Ty (sic) for sharing this, it’s very illuminating.”
Playground AI founder Suhail Doshi responded to Wang’s post, writing: “The models aren’t instructable like that so it’ll pick any generic thing based on the prompt. Unfortunately, they’re not smart enough.” He added that the company hopes to solve this issue moving forward.
Wang told Business Insider that her initial reaction was “amusement”, however, she was happy to prompt a debate around “AI bias.”
“However, I’m glad to see that this has catalysed a larger conversation around AI bias and who is or isn’t included in this new wave of technology.”
She added that “racial bias is a recurring issue in AI tools” and that, given her recent experience, she’s not likely to use them again.
Wang: “I haven’t gotten any usable results from AI photo generators or editors yet, so I’ll have to go without a new LinkedIn profile photo for now!”
The MIT student told The Globe that she hopes its creators, and other AI companies, rectify this issue: “I hope people who are making software are aware of these biases and thinking about ways to mitigate them.”
Many artists have been vocal in the past about the problems they’ve encountered with AI while trying to depict race in images.
Stephanie Dinkins who frequently combines art and technology, revealed that AI has often stereotyped or censored Black culture in her experience.
“The biases are embedded deep in these systems, so it becomes ingrained and automatic. If I’m working within a system that uses algorithmic ecosystems, then I want that system to know who Black people are in nuanced ways, so that we can feel better supported,” she told The New York Times.
Related links:
AI creates what ‘most attractive man’ looks like in every country
This is the world’s first artificial intelligence robot artist
AI used to create a ‘lost’ Oasis record (and it’s actually pretty good)
AI-generated woman successfully sells nudes to unsuspecting Reddit users
