Dr. Fei-Fei Li: The benevolent scientist
Dr. Fei-Fei Li:
The benevolent scientist
Fei-Fei Li, the internationally-celebrated scientist, has always been a champion of the positive power of artificial intelligence (A.I.).
At the new Stanford Institute for Human-Centered Artificial Intelligence, which specializes in research and education on the potential uses of machine learning, Li’s goal as co-director is to find ways in which A.I. can enhance people’s lives.
“A.I. has come of age. It has migrated from a niche technical field to a major driving force of our future in our society,” Li told “CNBC Meets: Defining Values.”
“We see that in industry, we see that in the daily conversations. And here at Stanford, we recognize that there's a need for thought leadership for where A.I. is going tomorrow and how it can be guided to make the maximum positive impact for human society.”
The idea for the new institute was born in 2016 as Li was standing on her driveway, talking to her neighbor John Etchemendy, Stanford’s former provost and philosopher.
They discussed the lack of diversity in A.I. and how the field was narrowly focused on engineering and algorithm. They decided there needed to be a broader discussion about how tech could help everyone.
Now, the institute gathers researchers and academics from different departments — such as economics, business, political science and computer science — to tackle subjects like the impact of A.I. on jobs and how people can re-skill.
“I think one challenge is to ensure that the future of this technology is benevolent,” Li told CNBC’s Tania Bryer.
“It’s here to collaborate, to augment, to enhance human lives and productivity and make everybody's life better. And related to that, is to democratize A.I. in a way that everybody gets benefit. Not just a few, or a selected group.”
One of the main potential benefits of A.I., according to Li, is in health care. Li has been working with Arnold Milstein, a professor of medicine at Stanford University, for the last seven years.
They’ve been researching how A.I., smart sensors and deep learning algorithms can be a force for good, from helping older people to lead healthier and more independent lives, to improving patient safety in intensive care units.
“Do you know that hospital-acquired infections kill three times as many people as car accidents a year in the United States?” Li said. “And that is mostly due to lack of hand hygiene practice in the hospital units. So, these are the kind of opportunities that we think a technology like A.I. can play a very positive role. Just like what it can do to transportation or self-driving cars.”
Born in Chengdu, China, Li’s family followed her father to live in the U.S. when she was 16, and settled in New Jersey.
Li worked odd jobs, such as in Chinese restaurants and in her parents’ dry-cleaning store — a situation she previously described as “survival mode.”
As a child Li remembers looking up at the night sky and wondering where the stars came from.
She was drawn to studying math and science at school, and she said the first place she visited in the U.S. was the campus at Princeton, where Albert Einstein spent the latter half of his life.
Years later, she would major in physics at the same college, although she did not have it easy as she had to travel home every weekend and work in the dry-cleaning store to fund her studies.
Li’s family went to the U.S. without much financial resources, she told CNBC. “So, the value of hard working, the value of staying strong as a family and love is the kind of support we have for each other, carried us through,” she said.
Li went on to study at Stanford and Caltech (The California Institute of Technology), and just like the famous 20th century physicists before her, Li loved to look “beyond the world of atoms” and ask questions about life.
“My own interest started turning into an equally foundational question about intelligence,” she said. “What is biological intelligence? What is human intelligence? And I started participating in neuroscience experiments. And then when I went to Caltech as a PhD student, I was researching in a neuroscience lab as well as in a computer science engineering lab studying the flip side of the question, which is, how do we make machines intelligent? So that's my road into A.I..”
Although Li is clear that intelligence and values can be instilled in a machine, she stresses that this is only possible due to the human behind it.
“When we design a technology, a piece of technology, when we make it into a product, into a service, when we ship it to consumers, the whole process included our values,” she told CNBC. “Or the values of the companies, or the solution makers. So by and large, we create technology to help people. To make lives better, to benefit people, to augment humans or to enhance humans. And I think from that point of view, A.I. is no different.”
Li said she was the only woman in her faculty for a long time — a “lonely process” — and although she says there has been an increase in the number of women majoring in computer science at Stanford, there is still a long way to go.
“Now we have a new generation of women professors. But this is far from enough,” she said. “This field has a severe lack of women and underrepresented minority and this will impact the technology. You look at all the face-recognition controversy, you'll see that from the training data, all the way to the kind of products that we create using a biased data, we have issues and we need to solve this problem at some point.”
There is a growing concern regarding the way that facial-recognition technology can misidentify women and people of color, as it has mostly been trained on white male faces, leading to incidents such as Apple being sued for $1 billion by a student claiming wrongful arrest.
One way to solve bias in the industry, Li said, is through educating and encouraging more women and minorities into the STEM (science, technology, engineering, and mathematics) sector.
In 2015 she launched Stanford Artificial Intelligence Laboratory's Outreach Summer Program (SAILORS), a three-week residential placement for mostly female high school students, and in 2017 this initiative evolved into AI4ALL.
Li said that just a few years later, she’s already seeing these students return to Stanford, starting their own outreach and leadership programs.
“Now we have an alumni base of more than 200 students of all kinds of backgrounds,” she said.
“Women, underrepresented, racial minority, low income family students and geographically diverse students. And we're seeing this message spreading. We're seeing their leadership spreading. And I think they're going to become a powerful force in A.I.'s future.”
Source: American Physical Society, 2017
Source: UNESCO Institute for Statistics, 2018
Artificial intelligence has been around now for 60 years, but people regularly voice concerns around issues such as technology getting into the wrong hands and robots outsmarting humans, a concept popularized by movies such as “Ex Machina.”
But A.I. still has a long road ahead to compete with human ability.
In 2015, Li gave a Ted Talk — viewed more than 2.5 million times — about a project she led called ImageNet which required downloading more than a billion images to train a computer to describe basic objects in a photograph. The result in computer science was revolutionary but the human equivalent, Li said, was a toddler uttering its first sentences.
However, Li does not dismiss real concerns around A.I. for the future.
“Every technology is a double-edged sword. So when humans discovered fire, it changed the lifestyle of the prehistoric humans. But it also can be dangerous. And every step of the way in our civilization, we've seen technology playing both very positive roles, as well as creating or introducing perils. And A.I. has that.”
Li said she was concerned about A.I. being used unfairly, whether that use was intentional or not, as well as regulation and privacy issues.
“How do we create an informed dialogue between policymakers and A.I. technologists so that we can together find solutions to critical issues like (the) future of jobs, distribution of resources, protecting our environment, education, and all these?”
Li is a pioneer for tackling these issues head on. Although the Institute for Human-Centered Artificial Intelligence only launched in March, there are plans to host a boot camp for policymakers to discuss regulation, as well as an executive education program for business leaders.
The institute also wants to encourage students to collaborate on ethics in computing and tech.
“We have regulated technology. And we will be regulating A.I. just like we regulated driving,” Li said. “We regulated nuclear technology and it's through policy, through regulation, through multinational cooperation. All these efforts are essential.”
She added: “So in this process, warnings are important. And we need thoughtful warnings. We need thoughtful preparation. So I don't reject that kind of negative warning message. But I want to focus on making a positive impact. Because this technology can save lives. This technology can create (a) better work environment or work process for many people. Can optimize our cities, can help combat climate changes. Can help our teachers in classrooms.”
So could machines one day be running amok on our planet?
Li is determined that is not going to be the case.
“I still believe the world is created by us,” she said. “And whatever future world we envision or we want to live in, is due to the work we do today. So, if we focus on human-centered A.I., human-centered technology, I hope that the future we create is human centered and it's benevolent.”
Writer: Rachael Revesz
Design and code: Bryn Bache
Editor: Matt Clinch
Series Presenter and Executive Producer: Tania Bryer
Series Executive Producer: Martin Conroy
Series Producer: Ged Cleugh
Series Associate Producer: Michelle Blackwell
Images: CNBC and Getty Images