Through her research on artificial intelligence and computer vision, Ngan Le works to bridge the gap between human intuition and machine logic. 

"Understanding the physical world is incredibly hard for computers, because the world is complex, dynamic and full of uncertainty," said Le, an assistant professor in the Department of Computer Science and Computer Engineering at the U of A.

Le's lab is teaching robots to interact with and move through unfamiliar spaces. She is training a robot arm to recognize an object regardless of its orientation. The robot arm can then be asked to grasp, push, pull, move or open the item. She is giving computers "object permanence," so they can keep track of someone even after that person disappears behind an obstacle. 

Le is also developing a computer system that sees things a person never could. In her work with the poultry industry, she is training a computer to weigh a chicken by sight, to look into the face of a newborn chick and know whether it is male or female, or continuously track, monitor and identify indicators of chicken welfare. The system can detect gait abnormalities and report round-the-clock behavioral patterns of individual chickens such as eating, drinking, dust bathing, sitting and walking. 

"Ultimately, what motivates me is the opportunity to create systems that don't just perform well in isolated tasks and controlled environments but integrate into the physical world in meaningful ways," she said. 

LOGIC AT THE CORE 

When Le arrived at college, she had never seen a computer. The machines seemed almost magical. But Le, who is from Vietnam, had always been drawn to math and logic. 

"Like all kids, I was really fascinated by problem-solving and creativity," she said. 

She realized that logic and imagination were the keys to computers. She soon caught up with her classmates who had been coding since middle school. 

For her undergraduate senior thesis, Le and another student developed software to help deaf children learn to speak. The computer was programmed to watch the person talk, then give them feedback on how to correctly move their mouth to form a word. The system received several national awards and was used in disability centers across Ho Chi Minh City, Vietnam. 

"I really love computer science, because the deeper you go, the more challenges you see," Le said. 

She came to the United States to complete a second master's degree and a Ph.D. at Carnegie Mellon University. In 2019, Le joined the faculty at the U of A. 

THE ETHICS OF AI 

For Le, research often begins outside the lab. She wants to understand the needs of industries, so her lab can find a solution. Too often, she said, researchers assume they know what industry wants and then create solutions for problems that do not exist. 

"It ensures that the solutions we develop are not just elegant on paper, but are robust, scalable and accessible in the real world," she said. 

She has received grants from National Science Foundation and the National Institutes of Health, along with support from companies including Tyson Foods, Cobb and Aviagen. U of A's Technology Ventures is pursuing  patent protection for several of Le's inventions to improve poultry processing. 

The ethics of artificial intelligence are also a major concern for Le. She wants to build tools that are fair, unbiased and aligned with society's values. Too often, AI systems are black boxes. Not even the people who build them can explain how they operate. Le believes that AI should be transparent. For example, her lab created a tool that mimics the gaze patterns of radiologists to read chest X-rays. Because the system uses a transparent methodology, doctors and patients can understand and trust its diagnosis. 

"What I'd like people to know is that AI isn't about flashy applications — it's a foundational technology that's enabling solutions to complex problems," Le said. "However, as it changes our lives, we need to be thoughtful about its design, deployment and governance."