"Everyone Has a Role to Play in Addressing AI Bias"
As more societies move up the curve of digital transformation and Artificial Intelligence (AI), people need to pay attention “to what they see and what they don’t see so that any hidden bias can be addressed," said Steve Crown, Vice President and Deputy General Counsel for Human Rights at Microsoft.
“AI is a human creation and people determine what data is fed into it, which can betray some selection and skewing,” said Crown, whose online Distinguished Lecture organized by UCA on March 16th 2021 focused on AI and Human Bias. This bias happens not necessarily because of intentional wrongdoing on the part of computer scientists who develop them, but oftentimes due to the numbers and quality of data they work with, said Crown. He cited the case of facial recognition technology, many of which are skewed towards those with lighter complexions, particularly men. There are more images of white men on mass media, “which then affects data sets for AI,” said Crown; data training sets are also biased for images with good focus and lighting.
Facial recognition technology is often skewed towards those with lighter complexions, particularly men, because of an overabundance of data training sets of white men on mass media, said Steve Crown during his lecture.
In every deployment of AI, it’s important to know what perspectives, assumptions, and biases affect performance because it matters, said Crown. AI can play hidden roles in matters such as allocation and withholding of resources and opportunities, in risk analysis for loan decisions, in hiring decisions, among others. “AI is the most powerful technology humans have created,” he said, adding that it is “exhilarating, but also humbling,” because there are still many unanswered questions. “It’s a challenge for the entire human community, and we all have a role to play.”
Microsoft is tackling the challenge of bias “from a broader ethical and human rights perspective,” and by focusing on responsible and “human-centered AI,” said Crown. Its foundational building blocks for AI are fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability.
Shaikh, one of Microsoft’s lead engineers behind Seeing AI, uses the app himself since he is blind.
Likening bias to gravity – which can be both useful and destructive – Crown said it can be harnessed by using AI to help solve the world’s problems. He cited Microsoft’s Seeing AI, a free app that helps the blind and low vision community navigate the world around them by describing people, objects and things they encounter. The app uses a smartphone’s camera to capture what’s happening around the user and narrates this through speech. Saqib Shaikh, one of Microsoft’s lead engineers behind Seeing AI, uses the app himself since he is blind.