The Stanford computer scientist addressed Yalies in a gripping conversation about human perspective and artificial intelligence.
Omar Ali, Contributing Photographer
At a Tanner lecture on Wednesday, renowned Stanford professor of computer science and co-director of the Human-Centered AI Institute Fei-Fei Li discussed the profound implications of humanistic artificial intelligence.
Li served as Vice President at Google and Senior Scientist in Artificial Intelligence/Machine Learning at Google Cloud while on sabbatical in the 2017-18 academic year. Her work focuses on the ethics of the development of algorithms and artificial intelligence. Li is also the creator of ImageNet, a pivotal project and extensive dataset that has contributed to recent advances in deep learning and AI.
During the lectureLi said that ImageNet has used neural network algorithms to succeed in object recognition in the field of computer vision.
“Some call it the fourth industrial revolution,” Li said of object recognition. “The field of computer vision has really blossomed as a result.”
Founded by Obert Clark Tanner and Grace Adams Tanner of the University of Utah, the Tanner Lectures on Human Values aim to instill intellectual promise and moral awareness—promoting courage and introspection of one’s place in society.
Lectures will be held at seven participating universities—Yale, Oxford, Stanford, Cambridge, University of California, Berkeley, University of Michigan, and University of Utah—during the 2021-22 and 2022-23 academic years. This event was public and organized in partnership with the Whitney Humanities Center.
“As humanists we are with the [Whitney Humanities Center] collaborated across campus, with a far wider reach than usual,” Diane B. Brown, associate director of the Whitney Humanities Center, told the news. “We personally invited lecturers and doctoral students from computer science, data science, psychology, philosophy and beyond. We designed this public lecture – and a private event tomorrow – to be truly transdisciplinary.”
According to Brown, psychology professor Marvin Chun had described Li as one of the most influential researchers in the history of AI.
In 2012, the winning algorithm of the ImageNet object classification challenge called AlexNet significantly reduced the generalization error and ushered in a new era in the deep learning revolution. ImageNet has been used in around 40,000 citations and provides universal data for image classification.
In her presentation, Li addressed how to prevent the looming impact of AI on society. People, she said, need to understand the true nature of non-artificial artificial intelligence: made by humans, designed to behave like humans, and therefore affect humans.
Li argued that visual perception and intelligence are deep and contextual – meaning AI must reflect these nuances with sensitivity. Li aims to make the fuzzier dimensions of human experience part of AI, noting that “relationships between objects need to be encoded.”
“We want to build technology that will help humanity,” Li told News.
Li described a motivation to improve the human condition through the constructive use of AI through exploring social and ethical dimensions in robust algorithms. During her talk, she carefully examined the intersection of visual and artificial intelligence that allows computers to imagine what humans see.
She also analyzed the problem of labor threats in the face of existing labor shortages, noting that “AI should augment human capabilities, not replace them”. Li also explained how AI can be used to improve the quality of life, including by relieving doctors working in congested emergency rooms.
Brown wrote that Li “eloquently demonstrated the human uses of AI” by explaining “how AI can complement, not replace, human work”.
“I was skeptical at first, but it’s interesting, when you study these things you learn more about yourself,” Robert Orr, a New Haven-area urbanist and participant, told the News. “It’s our quirks, our differences, and when they flow together, things start to happen.”
Tanner’s past instructors have included Elaine Scarry, Salman Rushdie, Oliver Stacks and Judith Butler.