AUSTIN (KXAN) — Researchers at the University of Texas at Austin are leading the way when it comes to fine-tuning the accuracy of artificial intelligence.
UT’s Machine Learning Lab held a public lecture Friday titled, “AI for Accurate and Fair Imaging.”
Researchers with the lab’s Institute for Foundations of Machine Learning (IFML) have been working to improve the algorithm that in 2020 produced an internet-famous image of former President Barack Obama, dubbed “White Obama.”
AI that was meant to enhance a pixelated, low-resolution photo of the 44th president instead transformed him into a white man.
“Even though it looked like a good image, a high resolution, realistic image of a person, it had a bias,” said Alex Dimakis, IFML co-director.
Dimakis and his team have been able to improve the technology with good results.
They dug into the initial data used to train the algorithm (made up of mostly white celebrities) but found that wasn’t the issue. The issue was the way the algorithm was built.
“The obsession with getting the right answer, it tends to amplify even a small bias in the data set,” Dimakis told KXAN.
“We saw, for example, a turban in one of the images,” he said, referring to another set of enhanced photos. “[The turban] was constructed like hair.”
“It can be very discriminatory,” said IFML Director Adam Klivans.
Klivans added advancements in this realm of AI could have benefits beyond race and gender; it could also help with medical imaging, giving doctors better images to review.
“If you have an MRI or a CAT scan, and it’s noisy,” he said, “this technology is excellent for that.”