‍

When Fei-Fei Li arrived in the United States as a teenager she faced enormous challenges and carried a huge weight of responsibility for her family’s well being. She went on to become one of the world’s most important and influential figures in the field of artificial intelligence, and is now using her considerable experience to shoulder an even greater burden of responsibility, guiding the future of AI to ensure that it works for the well being of humanity as a whole. 

‍

‍

The American Dream Come True

Fei-Fei was born in Beijing in 1976, and grew up in the Chinese city of Chengdu; her father was a repairman who specialized in cameras, while her mother worked as a cashier. She describes her childhood self as “nerdy”, with her head always buried in a book, and she was a promising high school student, but her life was soon to be turned upside down. Her father had moved to the United States in 1988 in search of work, and four years later the rest of the family followed him, settling in the small town of  Parsippany, New Jersey. 

‍

At first, none of the family spoke English, and the responsibility to act as translator fell on Fei-Fei’s shoulders as she attempted to adapt to her new environment. She credits a handful of kindly teachers at her new high school with helping her overcome those difficult first weeks and months, as she developed her language skills and began to recreate the excellent grades that she was used to getting in Chengdu. She was also forced to work extremely hard outside of school in low-paid jobs including a house cleaner and waitress at local Chinese restaurants in order to support her family. She has later described how isolated she and her family were during this period, saying ”We had some immigrant friends, but everyone was busy and we were just in survival mode. I didn't make a lot of friends in high school. It's a cruel time and I was very geeky.”

‍

‍

Nonetheless, she thrived in Parsippany’s education system, and was even featured in the local paper with her high school principal lauding her as an example of “The American Dream come true”, despite her own reservations that “I don’t feel I totally belong here still”. Life continued to be a struggle for Fei-Fei and her family even after she was accepted into Princeton to study physics, but rather than take on work to remedy this herself, she chose to take a larger risk, borrowing money from friends and teachers at the university to buy a dry cleaning store for her parents. To ensure her investment succeeded she would travel the 50 miles north to Parsippany most weekends to work in the store herself, while providing telephone support to her parents during the week. Fortunately, having to juggle her academic endeavors and business ventures did not hold Fei-Fei back, and she graduated with flying colors in 1999, before going on to complete her Ph.D. and eventually joining the faculty at Princeton’s computer science department as an assistant professor. 

‍

The Eyes Of The World

It was here that Fei-Fei began to pioneer the software that would lead to her being described as “one of a tiny group of scientists—a group perhaps small enough to fit around a kitchen table—who are responsible for AI’s recent remarkable advances” by Wired magazine. Her peers had been working on the problem of teaching computers how to recognise images, but were taking a piecemeal approach, training machines to recognise one type of object at a time. This was laborious and inefficient, and Fei-Fei wondered if a better approach could be taken by replicating the way children learn to process visual information, encountering and familiarizing themselves with a wide array of new objects at once. This thought led to the creation of ImageNet, an enormous database of images containing pictures of every type of object imaginable. Once the database was created, a new task presented itself which was also extremely laborious; humans would have to go through every image and tag each one with a description. With a little more than three million images in the dataset, this was an enormous task, and Fei-Fei struggled to find support for this undertaking within the Princeton faculty.

‍

Instead, she outsourced the task to the internet, using Amazon’s Mechanical Turk program to parcel the job out to an army of remote workers around the globe. Before long ImageNet was complete, and it has subsequently had an enormous impact on the development of artificial intelligence. The database has now grown to over fourteen million images, and is a fundamental part of countless AI applications that define the modern technological era, from captchas on webpages to self-driving car technology. The concept is also replicated in databases such as Stanford’s Medical ImageNet, which powers vital biotech applications for diagnosis and automated treatments.

‍

‍

Steering the future

Fei-Fei’s innovation has led to her becoming one of the most respected and influential figures in the field of artificial intelligence, but she has used that influence to make the case for responsible and ethical development of AI technology. The problem of bias within AI applications - where an AI system gives consistently different outputs for one group of people compared to another - is a particular concern for Fei-Fei, and something that she is seeking to reduce within ImageNet. She’s also working to ensure that new technologies being developed from her work are used to benefit humans as a whole, rather than divide them. In 2017 she founded the Human-Centered AI Institute at Stanford, with the stated mission of “Advancing AI research, education, policy, and practice to improve the human condition.” Soon afterwards, she took a sabbatical from Stanford to join Google as Vice President and Chief Scientist of AI and machine learning, and quickly became embroiled in a controversy that amply demonstrated the need for vigilance in this field.

‍

Soon after she joined Google, the firm secured a contract from the US Department of Defence for “Project Maven”, which sought to use image recognition techniques to improve the efficacy of Pentagon drone cameras. Google told employees that the project was “specifically scoped to be for non-offensive purposes”, and that “saving lives was the overarching intent”. This did not quell fears within the firm, and 4000 employees staged a protest against the contract, which was eventually dropped by Google a year later. Although not directly involved with the project, as the most recognizable name attached to it Li became the public face of the controversy, and drew heavy criticism for her involvement after internal emails featuring her advice to other Google executives were leaked to the New York Times. 

Li was concerned not only for her own reputation, but also the reputation of the field of artificial intelligence. To remedy this, she took the unusual step of prolonging her sabbatical at Google in order to rewrite the company’s ethics guidelines on artificial intelligence, before returning to Stanford. Subsequently, she has redoubled her efforts to promote the ethical use of AI, and also to ensure diversity among the people working in the field. One vital tool in this effort is AI4ALL, a non-profit organization she founded in 2015 which strives to promote the education of “historically excluded talent” - such as women and people of color -  within the industry. Through these initiatives Fei-Fei is leading the call to ensure ethical and responsible development of new AI technologies, and also working to increase the chances of achieving those results by ensuring that the people creating those technologies have a broad and diverse range of backgrounds. With the great power that artificial intelligence potentially wields, it is vital that the people shaping its future are aware of the great responsibility that comes with their work, and Fei-Fei Li’s dedication to this task singles her out as not just an example of the American Dream come true, but a True Superhero to boot.  

‍