What is The Worlds I See by Fei-Fei Li about?

The Worlds I See is a science memoir that chronicles Fei-Fei Li's journey from a teenage immigrant arriving in America with less than $20 to becoming a pioneering figure in artificial intelligence. The book interweaves her personal story with the rapid development of modern AI, including her groundbreaking creation of ImageNet, which revolutionized computer vision and deep learning. Li shares firsthand insights into AI's evolution and its implications for humanity's future.

Who is Fei-Fei Li and why is she called the "Godmother of AI"?

Fei-Fei Li is a Chinese-American computer scientist and Stanford professor known as the "Godmother of AI" for her groundbreaking work in computer vision. She invented ImageNet, a massive database of 15 million images that has been widely regarded as one of three driving forces behind the modern AI revolution. Li currently serves as Co-Director of Stanford's Human-Centered AI Institute and recently founded World Labs, focusing on spatial intelligence AI.

Who should read The Worlds I See by Fei-Fei Li?

The Worlds I See appeals to aspiring scientists, immigrants, technology professionals, and anyone interested in AI's development and ethical implications. Students and first-year professionals will find inspiration in Li's story of overcoming adversity through persistence and education. Policymakers, tech leaders, and general readers curious about how AI revolutionized our world and concerns about responsible technology use will gain valuable insights from her unique perspective at AI's forefront.

Is The Worlds I See by Fei-Fei Li worth reading?

The Worlds I See offers exceptional value as both an inspiring immigrant success story and an insider's account of AI's breakthrough moment. Princeton selected it as their 2024 pre-read program, with President Eisgruber praising how Li "beautifully illuminates the persistence that science demands". The book provides rare firsthand perspective from a central figure in AI development while addressing crucial questions about technology's ethical future, making it essential reading for understanding both AI's history and its human-centered potential.

What is ImageNet and why is it significant in The Worlds I See?

ImageNet is Fei-Fei Li's revolutionary database containing nearly 15 million images across 22,000 categories that transformed artificial intelligence. The 2012 ImageNet Challenge, where Geoffrey Hinton's team achieved breakthrough results, represents AI's defining moment that Li describes in her memoir. This dataset enabled rapid advances in computer vision throughout the 2010s and is widely considered one of three critical driving forces behind the birth of modern deep learning.

How does Fei-Fei Li's immigration story connect to AI development in The Worlds I See?

Li arrived in New Jersey at age 16 without English proficiency, money, or connections beyond her parents, working as a waitress and cleaner while attending high school. Her family rebuilt their lives from nothing, with Li helping run her parents' dry-cleaning business on weekends even while attending Princeton on full scholarship. This immigrant journey of resilience, adaptation, and persistence directly parallels AI's own evolution—both required embracing uncertainty, learning from limited data, and pushing through countless setbacks to achieve breakthrough moments.

What does The Worlds I See reveal about human-centered AI?

The Worlds I See advocates for AI development that prioritizes human impact, ethics, and inclusivity rather than technology for its own sake. Li co-founded AI4ALL in 2017 to increase diversity in AI education and has testified before Congress about responsible technology use. Her vision emphasizes that AI should be developed with diverse perspectives and serve humanity's needs, drawing from her experience advising the White House, serving on the UN Scientific Advisory Board, and leading Stanford's Human-Centered AI Institute.

What role did Princeton play in Fei-Fei Li's journey described in The Worlds I See?

Princeton provided Li with a transformative full scholarship that seemed so improbable she asked two advisors to verify the acceptance letter. She earned her bachelor's degree in physics in 1999 with High Honors while commuting home weekends to help her parents' dry-cleaning business. Princeton later recognized her achievements with Distinguished Alumni honors in 2020 and the prestigious Woodrow Wilson Award in 2024, selecting The Worlds I See as their university-wide pre-read program.

La science de nos mondes réunis

Curiosité, exploration et découverte à l'aube de l'IA

Fei-Fei Li

4.31(5308 Reviews)

Aperçu de La science de nos mondes réunis

Dans « The Worlds I See », la pionnière de l'IA Fei-Fei Li entremêle son parcours d'immigrée avec le développement de technologies révolutionnaires. Recommandé par Obama et choisi comme lecture préalable pour la promotion 2028 de Princeton, ce mémoire révèle comment la scientifique à l'origine d'ImageNet pose la question suivante : « L'IA peut-elle, en fin de compte, respecter la dignité humaine ? »

Choisissez la façon de le raconter

Un livre. De nombreuses façons de l'écouter.

Les mêmes idées de La science de nos mondes réunis. Choisissez la voix, la profondeur et le rythme qui vous conviennent. Changez à tout moment.

Chapitre 1

From Chengdu to Silicon Valley: A Pioneering Journey in AI

The hotel lobby felt strangely ordinary as I prepared to testify before Congress about artificial intelligence. Despite my exhaustion from a red-eye flight and constant rehearsing, I couldn't help reflecting on my first visit to D.C. as a teenage immigrant when the National Air and Space Museum sparked my love for science. Now, twenty years later, I'd found my way to the science of intelligent machines. In just five years, deep learning had transformed from an academic niche to a global force, bringing both excitement and growing concerns about biased algorithms, job displacement, and surveillance. My recent New York Times op-ed had attempted to balance optimism with legitimate concerns, arguing that AI's impact would ultimately depend on human motivations. As I approached the Rayburn House Office Building, personal worries weighed heavily-I'd left my ailing mother's hospital bedside to attend. But she had insisted I come, seeing her daughter address Congress just two decades after arriving in America. Inside the chamber, seeing my name printed simply as "DR. LI" amid the ornate surroundings, I resolved to share my central belief: that AI's development must explicitly center on human benefit, and that harnessing it safely would require much more than science alone.

Chapitre 2

A Childhood Shaped by Curiosity and Displacement

Under a leafy canopy during a childhood hiking trip in China, I gazed at stars while our guide pointed out constellations, including the romantic tale of Zhi nu and Niu lang separated by the "river of stars" - our galaxy. Even at ten, I felt deeply drawn to exploring nature's wonders, from plants and insects to distant mountains and celestial bodies. Born in Beijing but raised in Chengdu, my life was marked by displacement. My maternal grandfather, once a promising student forced to abandon education during wartime, passed down an unresolved tension - a feeling that home lay elsewhere. My father, an electrical engineer with an "allergy to seriousness," showed me curiosity in its purest form during our cycling adventures through countryside and city. Even my name "Fei-Fei" came from his bird-watching on the day of my birth, the Mandarin word *fei* meaning "to fly." My mother provided direction to this curiosity. Born to an intellectual lineage with a naturally fierce mind, her academic promise was thwarted during the Cultural Revolution due to her family's association with the Kuomintang. Her frustration manifested as a burning desire to escape, finding refuge in books that became windows to worlds beyond her reach-a passion she eagerly shared with me. Through Chinese translations of Western classics, I developed my own love for imagining alternate realities. Our family lived in a fourth-floor apartment in one of three identical Soviet-influenced towers on Chengdu's outermost ring road. My maternal grandparents, who lived nearby, were central to my childhood with their weekly lavish meals featuring coastal-style cooking that subtly reinforced our outsider identity in Chengdu. They nurtured my imagination with books on diverse topics from marine life to robotics to Chinese mythology, treating me as a person first and a girl second. My schooling in Chengdu was engaging, combining math and science with humanities like geography, poetry, and history. This contentment shattered one afternoon when our teacher dismissed the girls but kept the boys behind. Hiding nearby, I overheard her admonishing the boys for underperforming, stating they were "biologically smarter than girls" and assuring them that girls would "naturally grow stupider" as teenagers. These words from a female teacher ignited an unfamiliar anger in me-not discouragement but indignation. I began noticing other signs that my family's inclusivity wasn't universal. Rather than dissuading me, these experiences strengthened my resolve to reach further. In magnet middle school, I embraced being a "tomboy" as a personal mission-cutting my hair short, refusing dresses, and pursuing interests that defied expectations like aerospace science and hypersonic aircraft design.

Voyez chaque mot au fil de l'écoute

Suivez facilement la lecture tandis que chaque mot de La science de nos mondes réunis s'illumine en temps réel.

Chapitre 3

Revolution and Escape: The Journey to America

In 1989, everything changed. School schedules became erratic with unexplained breaks, and teachers' demeanors shifted with patriotic overtones woven into every subject. At home, my parents spoke in hushed, cryptic tones about something exciting on the horizon. Then suddenly, the cheer vanished, replaced by closed-door conversations from which I was deliberately excluded. Through late-night eavesdropping, I caught fragments mentioning "education," "opportunity," "freedom," and frequently, my name. Finally, my parents revealed that my father would be moving to America-the first step in a larger plan orchestrated primarily by my mother. He would secure work and housing, and we would join him shortly after. Within weeks, he was gone, taking a third of my familiar home life with him. By 1992, when I turned fifteen, our visas finally arrived. Our journey began with a trip to Shanghai, where I insisted we visit the Bund. I was particularly drawn to the Pujiang Hotel where Einstein had once stayed-finding comfort in remembering that he too had been an immigrant. Boarding the plane, I took inventory of what I had: grandparents I loved but would lose, a father I longed to see again despite lingering hurt, and a mother whose confidence was faltering. I didn't yet have a clear identity, but I had physics. The rest would be left to fate.

Chapitre 4

The Immigrant Experience: Survival and Discovery

Our plane's landing at JFK jolted me back to harsh reality. This wasn't an adventure but the abrupt end of my only familiar life. Our plan to reunite with my father at the gate dissolved into hours of anxious waiting in baggage claim. When he finally burst through the doors, frantic after his car broke down in a tunnel, we were too exhausted for the joyful reunion it should have been. Our cramped one-bedroom apartment in Parsippany, New Jersey, quickly dispelled any illusions of American grandeur. I slept in a narrow gap between kitchen and dining area, while we furnished our space with discarded items found on curbs. Just forty-eight hours after arrival, I faced my next challenge: American high school. My first days at Parsippany High School overwhelmed every sense. Everything was brighter, faster, and noisier than my world in China. Between classes, loud bells unleashed roaring crowds that flooded hallways with teenage energy, unlike the docile atmosphere I knew. At home, even simple homework assignments took hours as I consulted two enormous dictionaries. Meanwhile, my parents worked exhausting jobs-my father repairing cameras and my mother as a cashier in a gift shop, her intellectual talents ignored. Working in a Chinese restaurant kitchen became my first job-twelve hours daily at two dollars per hour, off the books in a sketchy neighborhood requiring nightly rides home. What truly drained me wasn't the long hours but the apparent hopelessness of immigrant life. Everyone around me worked relentlessly without escaping menial labor. I immersed myself in academics, seeking stability amid chaos. During one weekend outing to Princeton University, I encountered Albert Einstein's bronze memorial that stopped me in my tracks. Reading the inscription describing him as "a Nobel Laureate in physics, a philosopher, a humanitarian, an educator, and an immigrant" who had lived nearby until 1955, I felt a fever breaking within me. The monument rekindled my passion for physics that had been reduced to mere survival.

Profitez davantage de La science de nos mondes réunis

Recevez des idées chaque jour parmi plus de 100 000 leçons
Les mots s'illuminent au rythme de l'audio
Votre bibliothèque sauvegardée sur tous vos appareils

Chapitre 5

Finding Mentorship and Direction

Mr. Sabella's towering height and curt tone made his reputation for sternness instantly manifest when I approached his desk after receiving an 89.4 on my math test. Though initially frustrated by what seemed like dismissal, I began to understand he was challenging me to stand out rather than blend in. His classroom and office-unofficially named "the Math Lab"-became my refuge. Despite language barriers, our conversations evolved beyond mathematics into literature. Unlike my parents, whose expertise I'd outgrown, Mr. Sabella recognized potential in me that I hadn't seen myself. When the thin envelope from Princeton arrived that December afternoon, I assumed rejection until spotting "Yes!" in bold type. The enclosed financial aid document revealed I'd received not just acceptance but a nearly full scholarship. For my mother, who had spent her life on the wrong side of unbridgeable divides, it offered unprecedented hope. Princeton's beautiful campus often seemed dreamlike. My journey had taken me from China's middle class through American poverty, through language acquisition and green card approval. Despite dark moments like witnessing a fellow student beaten for not fitting in, I'd discovered a wealth of humanity along the way-a community that welcomed immigrants, teachers who encouraged students struggling with English, and an Ivy League education.

Chapitre 6

The Birth of a Scientific Passion

In a darkened lab in 1997, I watched as electrodes translated a cat's visual cortex activity into sound-what the cat saw, we heard. Though I'd entered Princeton as a sophomore physics undergrad, this encounter with the mysteries of the mind would soon consume me completely. By sophomore year, my relationship with physics began shifting. I wondered if I was drawn to physics itself or the spirit that motivated it-the courage to ask brazen questions about our world. I devoured biographies of Einstein, Feynman, and Bohr, noticing how many physics giants developed late-career interests in biology and life itself. Books like Schrodinger's "What Is Life?" and Hofstadter's "Godel, Escher, Bach" captivated me, suggesting the mind could be understood in discrete, mathematical terms. My first solo trip within the United States took me to California for an eight-week research position at Berkeley. I discovered our project wasn't studying the brain biologically but computationally-extending Hubel and Wiesel's work by analyzing how cats' brains responded to full-motion video. In this understaffed, underfunded but unconstrained lab environment, I took on responsibilities years before I was qualified. The brain's sophistication dwarfs our most powerful supercomputers, operating within cubic inches and powered by mere calories. Despite the relative simplicity of individual neurons, their 100-billion-fold replication with 10^11 connections creates something transcendent-matter becomes mind. Through painstaking work with microelectrodes and signal processing, we successfully reconstructed the movies projected before a cat's eyes using only intercepted brain signals. Berkeley transformed my perspective. While Princeton meant regimented life between classes and helping at my parents' dry-cleaning shop, Berkeley offered a vibrant reality where I felt like a scientist, not an immigrant. The mysteries of intelligence captivated me more than physics ever had-more expansive yet more intimate. After years of not belonging, I'd found my place.

Voyez chaque mot au fil de l'écoute

Suivez facilement la lecture tandis que chaque mot de La science de nos mondes réunis s'illumine en temps réel.

Chapitre 7

Vision as the Gateway to Intelligence

According to zoologist Andrew Parker's hypothesis, photosensitivity-the foundation of the modern eye-ignited the Cambrian explosion that transformed Earth's ecosystem in just ten million years. Beginning with light-sensitive opsin proteins that could transmit bioelectric signals, vision evolved rapidly: first a shallow divot to discern light direction, then deepening into a pinhole-camera-like aperture, finally developing lenses that increased clarity. As my Princeton graduation approached in 1999, I faced a pivotal decision between scientific ambitions and financial security. Wall Street firms aggressively recruited me with promises of perks and eye-watering salaries that could relieve our family's financial burdens. When I mentioned these opportunities to my mother during a quiet moment at the dry-cleaning shop, she cut through my equivocation with characteristic directness: "Fei-Fei, is it what you want?" When I admitted I wanted to be a scientist, she simply replied, "So what are we even talking about?" Her clarity settled it-I was going to grad school. During my Caltech visit, I encountered professors who challenged me intellectually in ways I'd never experienced. Christof Koch posed profound questions about consciousness and visual experience, while Pietro Perona brought an infectious intellectual curiosity to complex questions. Despite my struggles with English and self-doubt, I found myself drawn to these brash personalities because our shared passion for science created an intellectual equality that transcended language barriers and status differences. The human capacity for vision has long defied automation attempts, but the potential applications of machine vision are enormous-from environmental monitoring to medical diagnostics. The challenge lies in translating pixels into meaningful concepts, a task that reveals vision as not merely an application of intelligence but synonymous with it. Understanding vision means understanding ourselves, a journey to the most foundational layers of our experience.

Chapitre 8

Breakthrough Moments in Computer Vision

At Caltech, we studied the visual cortex as a pathway to machine intelligence, believing that understanding human perception was essential to creating artificial versions. Anne Treisman's "feature integration theory" provided an early foundation, suggesting vision begins with identifying small details before establishing relationships between them. When Pietro handed me Simon Thorpe's brief three-page article from Nature, "Speed of Processing in the Human Visual System," it upended everything I thought I understood. Using EEG measurements, Thorpe had discovered that humans could identify the contents of photographs flashed for just 27 milliseconds-roughly a honeybee's wingbeat. Even more remarkably, the brain showed recognition just 150 milliseconds after seeing an image-about the time it takes to blink. By my second year at Caltech, I'd reached a powerful conclusion: visual perception relies fundamentally on categorization. Rather than drowning in endless details of light and color, our vision transforms the world into discrete concepts we can understand and react to instantly. This ability-to see a tree not as patterns of green and blue but as an object with known properties-was how our ancestors survived chaos. Testing our algorithm with Pietro in 2003, I watched as it attempted to identify an airplane in a photograph-a task trivial for humans but challenging for machines. What made this thrilling wasn't just that it worked, but how: unlike traditional approaches requiring hundreds of airplane images, our algorithm had seen only one before. We'd shown it hundreds of unrelated images first-jungle cats, motorcycles, human faces-proving that exposure to visual diversity improved specific recognition later. We called this "one-shot learning," mimicking humans' ability to recognize new objects after just a single glimpse. During another lunch at Red Door Cafe, Pietro proposed creating an entirely new, larger dataset from scratch. When I suggested starting with ten image categories (compared to Caltech's existing seven), Pietro pushed for more ambition. Our negotiation escalated until Pietro dramatically declared, "Let's do a hundred, Fei-Fei." Though initially shocked by the workload this would entail, I grew excited at how our model might perform with such a resource.

Profitez davantage de La science de nos mondes réunis

Recevez des idées chaque jour parmi plus de 100 000 leçons
Les mots s'illuminent au rythme de l'audio
Votre bibliothèque sauvegardée sur tous vos appareils

Chapitre 9

Personal Struggles and Scientific Perseverance

My academic momentum was abruptly halted when my mother suffered congestive heart failure. After rushing her to a doctor, I faced a devastating reality: we needed to sell the dry-cleaning shop that had been our lifeline for seven years. I decided to move my parents across the country to join me in my tiny Pasadena dorm. Speaking with Mr. Sabella, I questioned whether I could continue my studies while supporting my family. The McKinsey interview went surprisingly well, and the company quickly offered me a permanent position with an attractive salary. I rushed home to tell my mother, conflicted between abandoning my scientific dreams and finally providing financial stability. But my mother was resolute: "We didn't come this far for you to give up now." She reminded me that my scientific journey had always been our family's collective mission since leaving Shanghai. Her words finally resonated-I would never question my path again. While trying to remember the word "beagle" in the lab one day, I had a revelation about our image dataset problem. My English dictionary provided the perfect solution-we could use the illustrated nouns throughout the book as an unbiased selection method for our categories. This immigrant's tool became instrumental in creating Caltech 101, the largest collection of images ever assembled for machine learning at that time: over 9,000 images across 100 categories. Our follow-up paper established a benchmark that researchers worldwide soon aimed to beat. Despite improvements with larger datasets like Caltech 101, I wondered if we needed even more categories. Searching the literature for answers, I rediscovered a paper by Irving Biederman from 1983 that suddenly appeared in a new light-his estimate that the world contained roughly 30,000 unique visual categories, not just 101. This revelation was staggering, suggesting our research had barely scratched the surface of what was needed to model human visual understanding.

Chapitre 10

The Birth of ImageNet

After my Princeton lecture, I met computational linguist Christiane Fellbaum. Though our fields seemed distantly related, we shared a fundamental interest in how minds conceptualize the world through categories-the intersection between vision and language. When she asked if I'd heard of WordNet, everything changed. Created by psychology legend George Miller, WordNet maps every concept humans articulate through language, organizing words by meaning rather than spelling. Christiane mentioned a related abandoned project called ImageNet that aimed to illustrate each WordNet concept with a single image-a name and concept that immediately resonated with me. The pieces suddenly aligned in my mind: WordNet capturing the world's concepts in a natural hierarchy, and the abandoned ImageNet attempting to match pictures to concepts. What if we created a dataset like Caltech 101 but scaled to WordNet's ambition? Despite seeming logistically impossible, the idea was powerful-not just in size but in diversity that reflects the world's true complexity. Despite my colleagues largely rejecting the concept of ImageNet, questioning the practicality of a dataset with thousands of categories, I persisted. Only Professor Kai Li, a Chinese immigrant like myself who understands exponential thinking, supported me by donating workstations. Together with brilliant student Jia Deng, we trimmed WordNet's 140,000 entries to 22,000 visual, countable objects, aiming for 1,000 photographs per category-a staggering 20 million images. When Amazon Mechanical Turk (AMT) entered the picture, everything changed. This crowdsourcing platform transformed our workforce from a handful of undergraduates to thousands of international contributors. Our completion timeline plummeted from nineteen years to under one year. By June 2009, the inaugural version of ImageNet was complete: fifteen million images across twenty-two thousand categories, annotated by forty-eight thousand contributors from 167 countries. However, when we presented at CVPR 2009 in Miami, the project received little attention, leaving me wondering if the naysayers were right about my hypothesis.

Voyez chaque mot au fil de l'écoute

Suivez facilement la lecture tandis que chaque mot de La science de nos mondes réunis s'illumine en temps réel.

Chapitre 11

Revolution in Machine Learning

Turning ImageNet into a competition required scientific rigor and careful design. We created extensive documentation explaining evaluation metrics, including a "top-5 error rate" allowing algorithms to provide five possible labels. We distributed a scaled-down subset-1.4 million photos covering a thousand categories-and created new test images with crowdsourced labeling. Despite initial enthusiasm, the results of the first "ImageNet Large Scale Visual Recognition Challenge" in 2010 were underwhelming with only incremental improvements. By 2011, participation dropped dramatically to just 15 entries. Our grand hypothesis that ImageNet would force revolutionary algorithm development seemed to be failing. By August 2012, after giving birth, I received an unexpected late-night call from Jia about the ImageNet Challenge results. An entry using a neural network-an approach many considered "ancient"-had achieved an astonishing 10-percentage-point leap in performance, setting a world record of about 85% accuracy. This breakthrough, called AlexNet after lead researcher Alex Krizhevsky, was steamrolling the competition. AlexNet was a convolutional neural network that mirrored biological vision, with layers of neurons learning their own sensitivities from training data. Though AlexNet built upon Yann LeCun's decades of work with CNNs, its key advantages were massive scale, affordable GPU hardware originally developed for video games, and most importantly, access to ImageNet's vast training data. The network contained 650,000 neurons with 630 million connections, trained continuously for a week until it could recognize a thousand different object categories with unprecedented accuracy. The Florence experience revealed how two different gambles had converged perfectly-my lab's yearslong pursuit of unprecedented data scale and Hinton's commitment to abandoned neural network algorithms. Though vindicated separately, their true power emerged only in combination. Flying home, I sat in contemplative silence, aware that history had just been made with only a handful of witnesses.

Chapitre 12

Human-Centered Artificial Intelligence

During a friend's bat mitzvah in summer 2013, I received an urgent call about my mother's health. While sitting with her during recovery, she surprisingly asked what I actually do for work. When my technical explanation about AI research didn't connect, I described practical applications like self-driving cars that could help someone with her condition. Her simple follow-up question-"What else can AI do to help people?"-became a pivotal moment, inspiring me to merge my AI expertise with healthcare needs. Dr. Arnie Milstein, a renowned Stanford medical professor, invited me to observe a demonstration of Philips' "eICU" technology-remote monitoring for intensive care units. During the presentation, I was stunned to learn that approximately 100,000 annual deaths result from preventable medical errors. Our conversation revealed awareness as healthcare's most precious and unscalable resource, connecting directly to my research on visual perception and attention. Following our eICU demonstration, Arnie and I assembled a diverse team to develop "ambient intelligence" systems for healthcare, focusing on monitoring hand hygiene in hospitals, tracking senior activities in care homes, and detecting patient immobility in ICUs. Despite Arnie's impressive ability to secure partnerships with real healthcare facilities, we faced a significant challenge: the scarcity of training data, as patients and clinicians are rarely recorded due to privacy concerns. After my mother's heart surgery, she refused to use her incentive spirometer despite its critical importance in preventing lung infections. When I finally confronted her at home, she revealed a profound truth: "Being a patient is horrible. It's not just the pain. It's the loss of control." This conversation crystallized the deepest lesson from my two years exploring AI in healthcare: human dignity transcends any algorithm or dataset.

Profitez davantage de La science de nos mondes réunis

Recevez des idées chaque jour parmi plus de 100 000 leçons
Les mots s'illuminent au rythme de l'audio
Votre bibliothèque sauvegardée sur tous vos appareils

Chapitre 13

The Future of AI and Personal Legacy

The Stanford Institute for Human-Centered Artificial Intelligence (Stanford HAI) now occupies a refurbished wing of the Gates Computer Science Building, symbolically placing humanistic values at the heart of computer science. HAI has become a hub for cross-disciplinary collaboration, bringing together law professors, political scientists, humanities scholars, and computational neuroscientists. Recent years brought alternatives to the human bottleneck in training models. Advances in model processing, size, parallelization, and "attention" mechanisms made it possible to train on data sets comprising significant portions of the internet itself. The resulting "large language models" (LLMs) demonstrated linguistic competence so fluent it was easy to forget one wasn't reading a human author's words. Scale was the operative word in this seemingly magical technology. While AlexNet debuted with sixty million parameters, transformers now reached hundreds of billions. The engineering challenges were endless, but the science surprisingly elegant-as if these possibilities had been waiting since LeCun's ZIP code reader or even Rosenblatt's perceptron. We just needed to make a simple idea big enough. On a bright afternoon, I watched my parents with their grandchildren in the gazebo's shade. My father, finally in his element as a grandfather, laughed and played with the children. A text from Stanford HAI's policy director announced the national research cloud had passed the senate. Another text from Jean Sabella showed her children excitedly opening Star Wars Lego sets from "Auntie Fei-Fei and Uncle Silvio." The Sabellas' compassion remains my greatest lesson in humanity. I find myself reflecting often these days. My parents' formative years come to mind-my mother locked away during the Cultural Revolution, my father lost in tragedy. I remember her trembling hands as we boarded the plane from China, and the dread at JFK when night fell without my father's arrival. Moving across the world marked me in ways I'm only beginning to understand. As a scientist and immigrant, "home" is never a clear concept. The best work happens at borders, where ideas exist between coming and going, explored by people who are both insiders and outsiders. This perspective gives us power to challenge the status quo. AI's future remains uncertain, with equal reasons for optimism and concern. What matters most is what motivates us as creators. As our field grows more diverse and inclusive, I'm more confident in our chances. In the mind, North Stars are limitless-each new pursuit beckons on its own horizon. My greatest joy comes from knowing this journey will never be complete.

★★★★★

Vous êtes arrivé au bout de La science de nos mondes réunis

Adopté par plus d'un million d'apprenants

“Cela fait 23 jours et je l'utilise tous les jours. Cette application fait désormais partie de ma routine.”

jayallen

Thèmes clés dans La science de nos mondes réunis

human-centered ai
immigrant scientist journey
computer vision evolution
gender bias in stem
academic pursuit of intelligence

Citations de La science de nos mondes réunis

AI's impact would ultimately depend on human motivations.
My father, an electrical engineer with an 'allergy to seriousness'.
I didn't yet have a clear identity, but I had physics.
I embraced being a 'tomboy' as a personal mission.
I resolved to share my central belief: that AI's development must explicitly center on human benefit.

Personnages de La science de nos mondes réunis

Fei-Fei LiThe author, an AI scientist and immigrant
Fei-Fei's motherAn intellectual woman who drove the family's move
Fei-Fei's fatherAn electrical engineer with a passion for nature
Fei-Fei's maternal grandfatherA former student who valued education

À propos de l'auteur

Le Dr Fei-Fei Li est une informaticienne pionnière, chercheuse en IA et auteure de « The Worlds I See: Curiosity, Exploration and Discovery at the Dawn of AI », un mémoire scientifique retraçant son parcours d'immigrée à l'une des figures les plus influentes de l'intelligence artificielle. Née à Pékin en 1976, Li est professeure Sequoia à l'Université de Stanford et codirectrice du Human-Centered AI Institute de Stanford.

Elle est surtout connue pour avoir créé ImageNet, la base de données révolutionnaire qui a transformé la vision par ordinateur et catalysé la révolution de l'apprentissage profond (deep learning). Ses recherches couvrent l'apprentissage automatique, la robotique et l'IA dans le domaine de la santé, avec plus de 300 articles scientifiques publiés.

Li a occupé le poste de vice-présidente chez Google et de scientifique en chef de l'IA/ML chez Google Cloud, et a cofondé AI4ALL, une organisation à but non lucratif promouvant la diversité dans l'enseignement de l'IA. Elle est actuellement cofondatrice et PDG de World Labs, qui développe une IA d'intelligence spatiale. Sa conférence TED de 2015 a été visionnée plus de 2 millions de fois. Elle a été nommée l'une des 100 personnes les plus influentes en IA par le magazine Time en 2023.

Telecharger le resume de La science de nos mondes réunis

Obtenez le resume de La science de nos mondes réunis en PDF ou EPUB gratuit. Imprimez-le ou lisez-le hors ligne a tout moment.

Les utilisateurs ont aussi aimé

AI 2041

Kai-Fu Lee & Chen Qiufan

AI Superpowers

Kai-Fu Lee

Couverture du livre AI Revolution: Promise, Peril, and Human Wisdom

6 sources

AI Revolution: Promise, Peril, and Human Wisdom

Verity Harding

Gregory Mone

Sierra Greer

Stuart Russell

Isaac Asimov

FAQ sur ce livre

Fait partie d'un plan d'apprentissage

PLAN D'APPRENTISSAGE

Master the Art of Small Talk

5 h 47 m•5 Épisodes

Découvrir plus

PLAN D'APPRENTISSAGE

The Culture: AI Minds and Human Purpose

This plan explores the intersection of advanced technology and social philosophy through the lens of Iain M. Banks's universe. It is ideal for sci-fi enthusiasts, futurists, and ethicists interested in the long-term implications of AI governance.

2 h•4 Sections

PLAN D'APPRENTISSAGE

Ai learning

As AI reshapes every industry, understanding its technical core and ethical boundaries is no longer optional. This plan is ideal for professionals and tech enthusiasts who want to transition from passive users to active creators of intelligent systems.

4 h 42 m•4 Sections

PLAN D'APPRENTISSAGE

AI's Impact on Society & Living in the AI Age

As AI rapidly reshapes our world, understanding its societal implications is no longer optional but a vital survival skill. This plan is designed for professionals and citizens who want to navigate the ethical, economic, and practical shifts of the digital age with confidence.

5 h 2 m•4 Sections

PLAN D'APPRENTISSAGE

The history and future of ai

As AI reshapes every industry, understanding its origins and technical mechanics is essential for informed decision-making. This plan is ideal for professionals and curious learners who want to move beyond the hype to understand the ethics and future of superintelligence.

5 h 32 m•4 Sections

PLAN D'APPRENTISSAGE

AI: weigh benefits & risks

As AI rapidly transforms every sector from healthcare to education, understanding its true potential and risks has become essential for informed citizenship and professional relevance. This learning plan equips anyone—whether business leaders, policymakers, students, or concerned citizens—with the critical thinking framework needed to navigate our AI-integrated future responsibly and effectively.

5 h 38 m•4 Sections

PLAN D'APPRENTISSAGE

Learn more about AI

As artificial intelligence reshapes every industry, understanding its technical and ethical foundations is no longer optional. This plan is ideal for professionals and students who want to move beyond the buzzwords to build actual systems while navigating the future of human-AI collaboration.

5 h 15 m•4 Sections

PLAN D'APPRENTISSAGE

AI Myths: LLMs vs. True Sentience

This learning plan is essential for anyone looking to look past the headlines and understand the actual capabilities of modern AI. It is particularly valuable for tech enthusiasts, students, and professionals who want to ground their understanding of machine intelligence in both science and philosophy.

5 h 45 m•4 Sections

PLAN D'APPRENTISSAGE

AI & social media algorithm

As AI and social media algorithms increasingly shape what we see, believe, and do online, understanding these systems is essential for anyone who wants to be digitally literate in the 21st century. This learning plan is ideal for professionals, students, creators, and everyday users who want to move beyond passive consumption to critically engage with the technologies that influence public discourse, personal behavior, and societal outcomes.

5 h 10 m•4 Sections

Cree par des anciens de Columbia University a San Francisco

BeFreed rassemble une communauté mondiale d'esprits curieux

4.7

note moyenne

Plus de 7,84 k notes sur l’app

BeFreed Community

"Instead of endless scrolling, I just hit play on BeFreed. It saves me so much time."

@Moemenn

"I never knew where to start with nonfiction—BeFreed’s book lists turned into podcasts gave me a clear path."

@Chloe, Solo founder, LA

117

"Perfect balance between learning and entertainment. Finished ‘Thinking, Fast and Slow’ on my commute this week."

@Raaaaaachelw

"Crazy how much I learned while walking the dog. BeFreed = small habits → big gains."

@Matt, YC alum

108

"Reading used to feel like a chore. Now it’s just part of my lifestyle."

@Erin, Investment Banking Associate , NYC

254

"Feels effortless compared to reading. I’ve finished 6 books this month already."

@djmikemoore

"BeFreed turned my guilty doomscrolling into something that feels productive and inspiring."

@Pitiful

4.5K

"BeFreed turned my commute into learning time. 20-min podcasts are perfect for finishing books I never had time for."

@SofiaP

"BeFreed replaced my podcast queue. Imagine Spotify for books — that’s it. 🙌"

@Jaded_Falcon

201

"It is great for me to learn something from the book without reading it."

@OojasSalunke

"The themed book list podcasts help me connect ideas across authors—like a guided audio journey."

@Leo, Law Student, UPenn

483

"Makes me feel smarter every time before going to work"

@Cashflowbubu

Decouvrez comment BeFreed est discute sur le web

Cree par des anciens de Columbia University a San Francisco

BeFreed rassemble une communauté mondiale d'esprits curieux

4.7

note moyenne

Plus de 7,84 k notes sur l’app

BeFreed Community

"Instead of endless scrolling, I just hit play on BeFreed. It saves me so much time."

@Moemenn

"I never knew where to start with nonfiction—BeFreed’s book lists turned into podcasts gave me a clear path."

@Chloe, Solo founder, LA

117

"Perfect balance between learning and entertainment. Finished ‘Thinking, Fast and Slow’ on my commute this week."

@Raaaaaachelw

"Crazy how much I learned while walking the dog. BeFreed = small habits → big gains."

@Matt, YC alum

108

"Reading used to feel like a chore. Now it’s just part of my lifestyle."

@Erin, Investment Banking Associate , NYC

254

"Feels effortless compared to reading. I’ve finished 6 books this month already."

@djmikemoore

"BeFreed turned my guilty doomscrolling into something that feels productive and inspiring."

@Pitiful

4.5K

"BeFreed turned my commute into learning time. 20-min podcasts are perfect for finishing books I never had time for."

@SofiaP

"BeFreed replaced my podcast queue. Imagine Spotify for books — that’s it. 🙌"

@Jaded_Falcon

201

"It is great for me to learn something from the book without reading it."

@OojasSalunke

"The themed book list podcasts help me connect ideas across authors—like a guided audio journey."

@Leo, Law Student, UPenn

483

"Makes me feel smarter every time before going to work"

@Cashflowbubu

"Instead of endless scrolling, I just hit play on BeFreed. It saves me so much time."

@Moemenn

"I never knew where to start with nonfiction—BeFreed’s book lists turned into podcasts gave me a clear path."

@Chloe, Solo founder, LA

117

"Perfect balance between learning and entertainment. Finished ‘Thinking, Fast and Slow’ on my commute this week."

@Raaaaaachelw

"Crazy how much I learned while walking the dog. BeFreed = small habits → big gains."

@Matt, YC alum

108

"Reading used to feel like a chore. Now it’s just part of my lifestyle."

@Erin, Investment Banking Associate , NYC

254

"Feels effortless compared to reading. I’ve finished 6 books this month already."

@djmikemoore

"BeFreed turned my guilty doomscrolling into something that feels productive and inspiring."

@Pitiful

4.5K

"BeFreed turned my commute into learning time. 20-min podcasts are perfect for finishing books I never had time for."

@SofiaP

"BeFreed replaced my podcast queue. Imagine Spotify for books — that’s it. 🙌"

@Jaded_Falcon

201

"It is great for me to learn something from the book without reading it."

@OojasSalunke

"The themed book list podcasts help me connect ideas across authors—like a guided audio journey."

@Leo, Law Student, UPenn

483

"Makes me feel smarter every time before going to work"

@Cashflowbubu

"Instead of endless scrolling, I just hit play on BeFreed. It saves me so much time."

@Moemenn

"I never knew where to start with nonfiction—BeFreed’s book lists turned into podcasts gave me a clear path."

@Chloe, Solo founder, LA

117

"Perfect balance between learning and entertainment. Finished ‘Thinking, Fast and Slow’ on my commute this week."

@Raaaaaachelw

"Crazy how much I learned while walking the dog. BeFreed = small habits → big gains."

@Matt, YC alum

108

"Reading used to feel like a chore. Now it’s just part of my lifestyle."

@Erin, Investment Banking Associate , NYC

254

"Feels effortless compared to reading. I’ve finished 6 books this month already."

@djmikemoore

"BeFreed turned my guilty doomscrolling into something that feels productive and inspiring."

@Pitiful

4.5K

"BeFreed turned my commute into learning time. 20-min podcasts are perfect for finishing books I never had time for."

@SofiaP

"BeFreed replaced my podcast queue. Imagine Spotify for books — that’s it. 🙌"

@Jaded_Falcon

201

"It is great for me to learn something from the book without reading it."

@OojasSalunke

"The themed book list podcasts help me connect ideas across authors—like a guided audio journey."

@Leo, Law Student, UPenn

483

"Makes me feel smarter every time before going to work"

@Cashflowbubu

Decouvrez comment BeFreed est discute sur le web

Commencez votre parcours d'apprentissage, maintenant

Chapitre 2

La science de nos mondes réunis

AperçuAperçu de La science de nos mondes réunis

Un livre. De nombreuses façons de l'écouter.

From Chengdu to Silicon Valley: A Pioneering Journey in AI

A Childhood Shaped by Curiosity and Displacement

Voyez chaque mot au fil de l'écoute

Revolution and Escape: The Journey to America

The Immigrant Experience: Survival and Discovery

Profitez davantage de La science de nos mondes réunis

Finding Mentorship and Direction

The Birth of a Scientific Passion

Voyez chaque mot au fil de l'écoute

Vision as the Gateway to Intelligence

Breakthrough Moments in Computer Vision

Profitez davantage de La science de nos mondes réunis

Personal Struggles and Scientific Perseverance

The Birth of ImageNet

Voyez chaque mot au fil de l'écoute

Revolution in Machine Learning

Human-Centered Artificial Intelligence

Profitez davantage de La science de nos mondes réunis

The Future of AI and Personal Legacy

Vous êtes arrivé au bout de La science de nos mondes réunis

Thèmes clés dans La science de nos mondes réunis

Citations de La science de nos mondes réunis

Personnages de La science de nos mondes réunis

À propos de l'auteur

Telecharger le resume de La science de nos mondes réunis

Les utilisateurs ont aussi aimé

FAQ sur ce livre

What is The Worlds I See by Fei-Fei Li about?

Who is Fei-Fei Li and why is she called the "Godmother of AI"?

Who should read The Worlds I See by Fei-Fei Li?

Is The Worlds I See by Fei-Fei Li worth reading?

Fait partie d'un plan d'apprentissage

Master the Art of Small Talk

Découvrir plus

BeFreed rassemble une communauté mondiale d'esprits curieux

BeFreed rassemble une communauté mondiale d'esprits curieux

Un livre. De nombreuses façons de l'écouter.

From Chengdu to Silicon Valley: A Pioneering Journey in AI

A Childhood Shaped by Curiosity and Displacement

Voyez chaque mot au fil de l'écoute

Revolution and Escape: The Journey to America

The Immigrant Experience: Survival and Discovery

Profitez davantage de La science de nos mondes réunis

Finding Mentorship and Direction

The Birth of a Scientific Passion

Voyez chaque mot au fil de l'écoute

Vision as the Gateway to Intelligence

Breakthrough Moments in Computer Vision

Profitez davantage de La science de nos mondes réunis

Personal Struggles and Scientific Perseverance

The Birth of ImageNet

Voyez chaque mot au fil de l'écoute

Revolution in Machine Learning

Human-Centered Artificial Intelligence

Profitez davantage de La science de nos mondes réunis

The Future of AI and Personal Legacy

Vous êtes arrivé au bout de La science de nos mondes réunis

Les utilisateurs ont aussi aimé

Fait partie d'un plan d'apprentissage

Master the Art of Small Talk

Découvrir plus

Aperçu de La science de nos mondes réunis