AI Superpowers

by Kai-Fu Lee

  It wasn’t just the memory of those fiery red blotches that made writing so difficult. My will had to be written in the traditional Chinese characters used in Taiwan—complex combinations of strokes, hooks, and flourishes far more intricate and elegant than the simplified characters used in mainland China. These characters constitute one of the oldest written languages still in use today, and I’d grown up immersed in it. I devoured epic kung-fu novels as a kid and even wrote one of my own when I was in elementary school.

  At the age of eleven I moved from Taiwan to Tennessee, a move inspired by my older brother, who was working in the United States and told my mother that Taiwan’s education system was too rigid and exam-oriented for a kid like me. It was tough for my mother to watch as her baby boy moved halfway around the world, and when we said goodbye, she made me promise one thing: that I would write her a letter in Chinese each week. In her letters back to me, she included a copy of the last letter I had sent to her, with corrections to those characters I had written wrong. That correspondence kept the written Chinese language alive for me as I went through high school, college, and graduate school in the United States.

  As I threw myself into a prestigious job at Apple in the early 1990s, our handwritten correspondence grew less frequent. When I moved to Beijing and began work with Microsoft, computers ate away more and more of the time I’d spent crafting traditional characters by hand. Writing Chinese on a computer was easier; it required typing out the romanized spelling of a Chinese word (for example, nihao) and then selecting the corresponding characters from a list. Artificial intelligence has further streamlined the process by predicting and automatically selecting the characters based on context. That technology has made typing Chinese almost as efficient as hammering out alphabetic languages like English.

  But gains in efficiency had turned into losses of memory. As I now sat hunched over the paper, I struggled to summon the shape of the characters after decades of neglect. I kept forgetting a dot or adding a horizontal stroke where it wasn’t meant to be. Each time I fudged a character, I would crumple up the paper and begin again.

  My will was just a page long, and in it I left everything to my wife, Shen-Ling. But my lawyer insisted that I write out four copies of that one page, each one to account for a different possible contingency. What if Shen-Ling died before me? Then I would give it all to my two daughters. What if one of them died? What if Shen-Ling and both of them died? It is an absurd set of hypotheticals to foist on someone grappling with his own mortality, but the law doesn’t carve out exceptions for a person’s internal distress.

  Those hypotheticals did, however, refocus my mind on what mattered. Not the management of my financial assets but the people in my life. Ever since I saw that PET scan, the world had seemed to dissolve into a whirlpool of despair, one with me at the center. Why did this happen to me? I’d never intentionally hurt anyone. I had always tried to make the world a better place, to create technologies that made life easier for people. I had used my fame in China to educate and inspire young people. I had done nothing to deserve dying at the age of fifty-three.

  Every one of those thoughts began with “I” and centered on self-righteous assertions of my own “objective” value. It wasn’t until I wrote down the names of my wife and daughters, character by character in black ink, that I snapped out of this egocentric wallowing and self-pity. The real tragedy wasn’t that I might not live much longer. It was that I had lived so long without generously sharing love with those so close to me.

  Seeing my ultimate end point threw my life into sharp focus and turned my egocentric wallowing inside out. I stopped asking why the world had done this to me, or lamenting that all my achievements couldn’t save me now. I began asking new questions: Why had I wanted so desperately to turn myself into a productivity machine? Why hadn’t I taken the time to share love with others? Why did I ignore the very essence that made me human?

  LIVING TOWARD DEATH

  As the sun set on Taipei, I sat alone at the table, looking at the four copies of my will, which had taken me four hours to write. My wife was in Beijing with our younger daughter, and I sat alone in the living room of my mother’s home. In the next room, my mother was lying down. She had for years suffered from dementia, and while she could still recognize her son, she had little ability to understand the world around her.

  For a moment, I felt grateful for the illness that clouded her mind—if she could understand the diagnosis that had just been delivered, I feared it would have broken her. She had given birth to me when she was forty-four, an age at which doctors urged her not to go through with the pregnancy. She refused to entertain that idea, seeing the pregnancy through and then showering me with endless affection. I was her baby, and she loved nothing more than feeding me her handmade spicy Sichuan dumplings, delicately wrapped bundles of pork that practically melted on your tongue.

  When I made the move to Tennessee, despite not speaking a word of English, my mother came and stayed with me for my first six months in America, just to make sure I was all right. Preparing to return home to Taiwan, she asked only that I continue to write her those letters in Chinese each week, a way to keep me close to her heart and rooted in the culture of my ancestors.

  She was someone who had spent her whole life sharing love with her children. Sitting at her dining table while she lay in the next room, I was racked by wave after wave of remorse. How had I been raised by such an emotionally generous woman and yet lived my life so focused on myself? Why had I never told my father that I loved him? Or truly shown the depth of caring for my mother before the dementia took hold?

  The hardest thing about facing death isn’t the experiences you won’t get to have. It’s the ones you can’t have back. Palliative care nurse and author Bronnie Ware has written extensively on the most common regrets that her terminally ill patients expressed in their final weeks of life. Facing the ultimate, these patients were able to look back on their lives with a clarity that escapes those of us absorbed in our daily grind. They spoke of the pain of not having lived a life true to themselves, the regret at having focused so obsessively on their work, and the realization that it’s the people in your life who give it true meaning. None of these people looked back on their lives wishing they had worked harder, but many of them found themselves wishing they had spent more time with the ones they loved.

  “It all comes down to love and relationships in the end,” Ware wrote in the blog post that launched her book. “That is all that remains in the final weeks: love and relationships.”

  Sitting at my mother’s table, this simple truth now burned within me. My mind swam backward through time, dipping in and out of memories of my daughters, my wife, and my parents. I hadn’t ignored the relationships in my life; on the contrary, I had very precisely accounted for each one. I had quantified them all and calculated the optimal allocation of time needed to achieve my objectives. Now I felt a gaping sense of emptiness, of irretrievable loss, about how little time for loved ones my mental algorithm had deemed “optimal.” This algorithmic way of thinking wasn’t just “suboptimal” at allocating time. It was robbing me of my own humanity.

  THE MASTER ON THE MOUNTAIN

  Like any epiphany worth having, these thoughts took time to truly sink in. I had felt something shift within me, but it would require patience and brutally honest self-examination to turn these pangs of regret into a new way of engaging with the world around me.

  Soon after my diagnosis, a friend recommended I visit the Fo Guang Shan Buddhist monastery in the south of Taiwan. Venerable Master Hsing Yun, a rotund monk with a soft smile, founded Fo Guang Shan in 1967 and remains at the monastery today. His monastic order practices what is called “humanistic Buddhism,” a modern approach to the faith that seeks to integrate core practices and precepts into our daily lives. Its monks eschew the stern mysteriousness of traditional Buddhism, instead embracing life with unconcealed joy. The monastery welcomes visitors from all backgrounds, sharing wit
h them simple practices and gentle wisdom. Around the monastery, you see couples getting married, monks enjoying a good laugh, and tourists taking a moment out of their busy lives to bask in the calm exuding from the people there.

  I had practiced Christianity while growing up in the United States, and although I no longer ascribe to a religious faith, I maintain a belief in a creator of this world and a power greater than our own. In visiting the monastery, I didn’t have any particular ambition—just a desire to spend a few days meditating on what I was experiencing, and reflecting on the life I had lived.

  One day after early morning classes, I was asked to join Master Hsing Yun for a vegetarian breakfast. The sun had not yet risen as we ate multigrain bread, tofu, and porridge. Master Hsing Yun now uses a wheelchair to get around, but his mind remains clear and sharp. Partway through our meal, he turned to me with a blunt question.

  “Kai-Fu, have you ever thought about what your goal is in life?”

  Without thinking, I reflexively gave him the answer I had given to myself and others for decades: “To maximize my impact and change the world.”

  Speaking those words, I felt the burning embarrassment that comes when we expose our naked ambitions to others. The feeling was magnified by the silence emanating from the monk across the table. But my answer was an honest one. This quest to maximize my impact was like a tumor that had always lived inside of me, ever tenacious and always growing. I had read widely in philosophy and religious texts, but for decades had never critically examined or doubted this core motivating belief within me.

  For a moment, Master Hsing Yun said nothing, using a piece of bread to wipe the last scraps of breakfast from his wooden bowl. I shifted uncomfortably in my seat.

  “What does it really mean to ‘maximize impact’?” he began. “When people speak in this way, it’s often nothing but a thin disguise for ego, for vanity. If you truly look within yourself, can you say for sure that what motivates you is not ego? It’s a question you must ask your own heart, and whatever you do, don’t try to lie to yourself.”

  My mind raced with rebuttals. I searched for the airtight logic that would redeem my actions. The days since my diagnosis had been an agonizing exercise in regret about the way I had engaged with my family and friends. I was slowly coming to terms with the emptiness of my emotional life. But as described in Elisabeth Kübler-Ross’s theory of the five stages of grief, before acceptance comes bargaining.

  Internally, I’d been trying to use my impact on millions of young Chinese people as a bargaining chip, as a way to balance out the lack of love shared with family and friends. I had over 50 million followers on Weibo, and I had relentlessly maximized my impact on this group. I even went so far as to build an AI algorithm for discovering and determining what other Weibo messages I should repost, always looking to maximize impact. Yes, I may have skipped out on family time to make public speeches, but think of all the people I had reached. I’d influenced millions of young students and tried to help a once-great country pull itself out of poverty. If you added it all up, wouldn’t you say that the good outweighed the bad? Couldn’t the gifts I’d given to so many strangers through my work make up for the dearth of love I had shared with those closest to me? Didn’t the equation balance out in the end?

  Now Master Hsing Yun was kicking the proverbial last leg of the stool out from under me. I tried to explain myself and cast my actions in the best light, based on what they had achieved. But he wasn’t interested in the results that my personal well-designed algorithm spat out. He patiently peeled away my layers of excuses and obfuscation. He continually directed the conversation inward, asking me to confront myself with unflinching honesty.

  “Kai-Fu, humans aren’t meant to think this way. This constant calculating, this quantification of everything, it eats away at what’s really inside of us and what exists between us. It suffocates the one thing that gives us true life: love.”

  “I’m just starting to understand that, Master Hsing Yun,” I said, lowering my head, staring at the floor between my two feet.

  “Many people understand it,” he continued, “but it’s much harder to live it. For that we must humble ourselves. We have to feel in our bones just how small we are, and we must recognize that there’s nothing greater or more valuable in this world than a simple act of sharing love with others. If we start from there, the rest will begin to fall into place. It’s the only way that we can truly become ourselves.”

  With that, he said goodbye and turned his wheelchair around. I was left with his words echoing in my mind and sinking into my skin. The time since my diagnosis had been a whirlwind of pain, regret, revelation, and doubt. I had come to understand how personally destructive my old ways of thinking had been, and I struggled to replace them with a new way of being human in the world that didn’t mimic some aspect of that algorithmic thinking.

  In the presence of Master Hsing Yun, I had felt something new. It wasn’t so much the answer to a riddle or the solution to a problem. Instead, it was a disposition, a way of understanding oneself and encountering the world that didn’t boil down to inputs, outputs, and optimizations.

  During my time as a researcher, I had stood on the absolute frontier of human knowledge about artificial intelligence, but I had never been further from a genuine understanding of other human beings or myself. That kind of understanding couldn’t be coaxed out of a cleverly constructed algorithm. Rather, it required an unflinching look into the mirror of death and an embrace of that which separated me from the machines that I built: the possibility of love.

  SECOND OPINIONS AND SECOND CHANCES

  While I wrestled with these stark realizations, the treatment for my cancer proceeded. My first doctor classified the disease as stage IV, the cancer’s most advanced stage. On average, patients with fourth-stage lymphoma of my type have around a 50 percent shot of surviving the next five years. I wanted to get a second opinion before beginning treatment, and a friend of mine arranged for me to consult his family doctor, the top hematology practitioner in Taiwan.

  It would be a week before I could see that doctor, and in the meantime I continued to conduct my own research on the disease. In my emotional life, I was turning away from the relentless pursuit of quantification and optimization. But as a trained scientist whose life hung in the balance, I couldn’t help trying to better understand the disease and quantify my chances of survival. Scouring the internet, I devoured all the information I could find about lymphoma: possible causes, cutting-edge treatments, and long-term survival rates. Through my reading, I came to understand how doctors classify the various stages of lymphoma.

  Medical textbooks use the concept of “stages” to describe how advanced cancerous tumors are, with later stages generally corresponding to lower survival rates. In lymphoma, the stage has traditionally been assigned on the basis of a few straightforward characteristics: Has the cancer affected more than one lymph node? Are the cancerous lymph nodes both above and below the diaphragm (the bottom of the rib cage)? Is the cancer found in organs outside the lymphatic system or in the patient’s bone marrow? Traditionally, each answer of “yes” to one of the above questions bumps the diagnosis up a stage. The fact that my lymphoma had affected over twenty sites, had spread above and below my diaphragm, and had entered an organ outside the lymphatic system meant that I was automatically categorized as a stage IV patient.

  But what I didn’t know at the time of diagnosis was that this crude method of staging has more to do with what medical students can memorize than what modern medicine can cure.

  Ranking stages based on such simple characteristics of a complex disease is a classic example of the human need to base decisions on “strong features.” Humans are extremely limited in their ability to discern correlations between variables, so we look for guidance in a handful of the most obvious signifiers. In making bank loans, for example, these “strong features” include the borrower’s income, the value of the home, and the credit score. In lymphoma s
taging, they simply include the number and location of the tumors.

  These so-called strong features really don’t represent the most accurate tools for making a nuanced prognosis, but they’re simple enough for a medical system in which knowledge must be passed down, stored, and retrieved in the brains of human doctors. Medical research has since identified dozens of other characteristics of lymphoma cases that make for better predictors of five-year survival in patients. But memorizing the complex correlations and precise probabilities of all these predictors is more than even the best medical students can handle. As a result, most doctors don’t usually incorporate these other predictors into their own staging decisions.

  In the depths of my own research, I found a research paper that did quantify the predictive power of these alternate metrics. The paper is from a team of researchers at the University of Modena and Reggio Emilia in Italy, and it analyzed fifteen different variables, identifying the five features that, considered together, most strongly correlated to five-year survival. These features included some traditional measures (such as bone marrow involvement) but also less intuitive measures (are any tumors over 6 cm in diameter? Are hemoglobin levels below 12 grams per deciliter? Is the patient over 60?). The paper then provides average survival rates based on how many of those features a patient exhibited.

  To someone trained in artificial intelligence—where even simple algorithms base decisions on hundreds if not thousands of distinct features—this new decision rubric still seemed far from rigorous. It sought to boil down a complex system to just a few features that humans could process. But it also showed that the standard staging metrics were very poor predictors of outcomes and had been created largely to give medical students something they could easily memorize and regurgitate on their tests. The new rubric was far more data-driven, and I leaped at the chance to quantify my own illness by it.