Public debate about artificial intelligence has intensified over the past six months with an outpouring of opinions about the dangers and ethics of a science that’s experiencing an exponential period of progress.
One of the key figures in this process, collaborating in each science and debate, is Fei-Fei LiSequoia Professor of Computer Science at Stanford University and Co-Director AI4Alla nonprofit organization promoting diversity and inclusion in the field of artificial intelligence.
Except for one controversy during her tenure as chief scientist at Google Cloud involving a proposal Google and Pentagon partnershipLi was something of a role model, not least due to her position in a field dominated by alpha male personalities.
Free from the influence of stylists and image creators, in interviews you come across with the skill of somebody who prefers to think through ideas as they arise slightly than making platform statements.
Li describes The worlds I see as a “double helix memoir.” One of the themes is the maturation of the science of artificial intelligence; the second is an account of her own growth as a scientist. According to her, the personal dimension got here to the fore after the initially “very nerdy book” was negatively assessed by a colleague.
Matter becomes mind
ITU photos, via Wikimedia Commons, CC BY
The story begins in Chengdu in China’s Sichuan province. As the only child in a family “in a state of quiet shock,” Li felt that her elders had experienced greater than they may appreciate. Her maternal grandparents, each academically educated, found themselves on the incorrect side of history during the Cultural Revolution. Her mother’s mental energy was thwarted.
As if there was an intertwined version of Yin and Yang in her heritage, her father’s free-spirited personality provided a complementary, if opposing, type of influence. According to Li, he was the variety of parent a child could create if left to his own devices. He was guided by his impulses and had various fascinations that took him on trips to rice fields in search of butterflies, stick insects and wild rodents.
Meanwhile, her mother was determined to flee. This ambition was realized in 1992 when the family moved to the United States. They lived in Parsippany, New Jersey, where 15-year-old Li, combating the demands of highschool in a foreign language, demonstrated her ability to work long hours toward the academic goals her mother valued.
My father’s fascination with natural life forms spilled over into the objective world of garage sales. He continued to have interaction Li in the practice of “investigating everything in sight.”
In “The Worlds I See,” Li reflects on the impact these two parents had on her academic profession. Without her mother’s dogged mental determination, she wouldn’t have been capable of proceed her highschool education given the family’s constant struggle for economic survival. Without her father’s childlike ability to pay complete attention to random phenomena, her research might never have found an revolutionary path.
The confluence of fascination and mental drive twists in unexpected ways. Ultimately, that is coupled with an almost visionary belief in what Li calls the North Star of her life: a calling to vary the parameters of understanding by asking “bold questions” of the kind asked by the great physicists who encourage her: Albert Einstein, Roger Penrose, Erwin Schrodinger.
Her own daring query – “what is vision?” – he focused steadily. For someone who likes to explain her endeavor in terms of revelation and revolution, her actual vision research doesn’t seem visionary in any respect.
An undergraduate degree in physics and computational mathematics at Princeton led to a summer job as a research assistant in the neuroscience group at the University of California, Berkeley. They tried to capture the cat’s neural responses to visual stimuli. The goal brain region was probed with capillary electrodes to detect signals.
These signals were first translated into sound waves after which back into visual patterns, from which the team was capable of recompose something near the original image shown to the animal.
Romance is difficult to return by, and yet Li comments, “Something transcendent is happening. Matter somehow becomes mind.”
What is data?
This knowledge sustains Li during her prolonged work. He becomes convinced that this principle will be applied to machine learning.
Following evidence that visual recognition in the human brain progresses from general to increasingly specific (bird, waterfowl, duck, mallard), Li and her postgraduate colleague decided to feed the computer with an intensive range of examples in a limited set of categories.
New imaging technologies from other fields have come to the rescue. Street View on Google identified 2,657 automobile models on the road in 2014. Amazon Clockwork Turk they escalated the scale and pace of their research as the categories multiplied, from the original ten to thirty, 100, one thousand.
But the project carried all the burdens that awaited Charles Darwin attempting to compile a comprehensive taxonomy of pigeons or James Murray compiling the Oxford English Dictionary.
For Li, the seemingly trivial belief that learning ought to be data-driven, not algorithm-driven, comes as a “moment of epiphany.” Bold questions: “What is vision?” and “what is intelligence?” to mix. They raise a third query: “what is data?”
A sudden thaw in “AI winter” of the first decade of the twenty first century began in 2012, when machine learning research made a breakthrough towards “big data”. The idea was to scale up and enhance the retention capabilities of AI to account for the scope and complexity of phenomena in the world itself.
Li stated that her approach is consistent with that Geoffrey Hinton, a Toronto-based cognitive psychologist who is usually credited with leading the artificial intelligence paradigm shift. Data will be multiplied exponentially, Hinton proposed, when machines refer to one another. Digital agents scan different areas of information and exchange the acquired knowledge to generate more sophisticated correlation modes.
Intelligence got here to be seen not as an innate property of a machine or the human brain, but as something that exists. It arises from interactions between objects, events, beings and environments. His development is more of a gatherer than a hunter.
Distributed intelligence
Distributed intelligence means distributed opportunities to participate in the joint evolution of human and machine intelligence. Great science and advanced technology are not any longer the exclusive domain of specialists whose ways of acquiring knowledge transcend the understanding of bizarre people. Everyone who has had an exchange from Chat GPT to Open AI is contributing.
However, Li emphasizes that effective human learning requires education. The most significant figure in her education was her highschool math teacher, Bob Sabella, who kept her on target as she struggled with the English curriculum. He remained a friend and mentor throughout her academic advancement.
Li says the true symbol of the way forward for human technology is a dedicated teacher. She co-founded AI4All in 2017, which goals to supply hands-on training to highschool students, especially girls, students of color, and people from immigrant families or low-income communities. Li herself matches most of those categories.
Youtube
The experiences Li describes in “The Worlds I See” reveal a unprecedented ability to persevere in the face of obstacles. She graduated from highschool, supplementing the family income by working at a Chinese restaurant for $2 an hour. While still a student, she ran a family dry cleansing business.
Her exams at Princeton were taken by special order at the hospital clinic where her mother was undergoing surgery for worsening heart problems.
But plainly every thing he experiences is returned to him in his pursuit of the North Star. Her mother’s recurring health problems introduced her to hospitals, which led her to research how artificial intelligence may very well be used, not to interchange the vital role of human nurses and health care employees, but to support them.
If Li’s efforts will be seen as a feminist endeavor, it is probably because the field in which she works is dominated by male celebrities who stubbornly see the future as a Darwinian struggle between human and machine intelligence.
“Which is smarter?” is a less audacious query than one which ought to be consigned to the dustbin of history. Speaking at a 2018 congressional hearing on power and responsibility in the use of advanced technologies, Li said:
There is nothing artificial about artificial intelligence. It is inspired by people, is created by people and, most significantly, influences people.
He clearly distances himself from those like Hintonwho see the current breakthrough in AI’s potential as an existential crisis, Li is interested in tangible societal threats and concrete ways to deal with them.
During a recent discussion with former US Secretary of State Condoleezza RiceLi, now director of the Hoover Institution at Stanford University, said she believes policy intervention can provide essential safeguards in areas where AI’s impact is more likely to be biggest.
These include its useful potential for health and education, in addition to the dangers of disinformation, lack of privacy and substitute of human labor.
If there may be an overarching theme in The Worlds I See, it’s that human and artificial intelligence form a double helix. How this may evolve and with what consequences will depend, says Li, on whether we create a “healthy ecosystem” in which talent, technology and public sector participation are coordinated.
