AI Superpowers

by Kai-Fu Lee

  7

  ★

  The Wisdom of Cancer

  The profound questions raised by our AI future—questions about the relationship among work, value, and what it means to be human—hit close to home for me.

  For most of my adult life, I have been driven by an almost fanatical work ethic. I gave nearly all my time and energy to my job, leaving very little for family or friends. My sense of self-worth was derived from my achievements at work, from my ability to create economic value and to expand my own influence in the world.

  I had spent my research career working to build ever more powerful artificial intelligence algorithms. In doing this, I came to view my own life as a kind of optimization algorithm with a clear goals: maximize personal influence and minimize anything that doesn’t contribute to that goal. I sought to quantify everything in my life, balancing these “inputs” and fine-tuning the algorithm.

  I didn’t entirely neglect my wife or daughters, but I always sought to spend just enough time with them so they didn’t complain. As soon as I felt I had met that bar, I would race back to work, answering emails, launching products, funding companies, and making speeches. Even in the depths of sleep, my body would naturally wake itself up twice each night—at 2 a.m. and 5 a.m.—to reply to emails from the United States.

  That obsessive dedication to work did not go unrewarded. I became one of the top AI researchers in the world, founded the best computer science research institute in Asia, started Google China, created my own successful venture-capital fund, wrote multiple best-selling books in Chinese, and amassed one of the largest social media followings in China. By any objective metric, my so-called personal algorithm was a smashing success.

  And then things came to a grinding halt.

  In September 2013, I was diagnosed with stage IV lymphoma. In an instant, my world of mental algorithms and personal achievements came crashing down. None of those things could save me now, or give me comfort and a sense of meaning. Like so many people forced to suddenly face their own mortality, I was filled with fear for my future and with a deep, soul-aching regret over the way I had lived my life.

  Year after year, I had ignored the opportunity to spend time and share love with the people closest to me. My family had given me nothing but warmth and love, and I had responded to that on the basis of cold calculations. In effect, mesmerized by my quest to create machines that thought like people, I had turned into a person that thought like a machine.

  My cancer would go into remission, sparing my life, but the epiphanies sparked by this personal confrontation with death have stuck with me. They’ve led me to reshuffle my priorities and to totally change my life. I spend far more time with my wife and daughters, and moved to be closer to my aging mother. I have dramatically cut down my presence on social media, pouring that time into meeting with and trying to help young people who reach out to me. I’ve asked for forgiveness from those I have wronged and sought to be a kinder and more empathetic coworker. Most of all, I’ve stopped viewing my life as an algorithm that optimizes for influence. Instead, I try to spend my energy doing the one thing I’ve found that truly brings meaning to a person’s life: sharing love with those around us.

  This near-death experience also gave me a new vision for how humans can coexist with artificial intelligence. Yes, this technology will both create enormous economic value and destroy an astounding number of jobs. If we remain trapped in a mindset that equates our economic value with our worth as human beings, this transition to the age of AI will devastate our societies and wreak havoc on our individual psychologies.

  But there is another path, an opportunity to use artificial intelligence to double down on what makes us truly human. This path won’t be easy, but I believe it represents our best hope of not just surviving in the age of AI but actually thriving. It’s a journey that I’ve taken in my own life, one that turned my focus from machines back to people, and from intelligence back to love.

  DECEMBER 16, 1991

  The routinized chaos of childbirth swirled all around me. Nurses and doctors in sanitary scrubs streamed in and out of the room, checking measurements and swapping out IV drips. My wife, Shen-Ling, lay on the hospital bed, fighting through the most physically and mentally draining act that a human being can perform: bringing another human into the world. It was December 16, 1991, and I was about to become a father for the first time.

  Our attending doctor told me it was going to be a complex labor because the baby was in the sunny-side up position, with her head facing toward the belly instead of toward the back. That meant Shen-Ling might require a cesarean section. I paced the room anxiously, even more on edge than most expectant fathers on the big day. I was worried about Shen-Ling and the baby’s health, but my mind wasn’t entirely in that delivery room.

  That’s because this was the day I was scheduled to deliver a presentation to John Sculley, my CEO at Apple and one of the most powerful men in the technology world. A year earlier, I had joined Apple as the chief scientist for speech recognition, and this presentation was my chance to win Sculley’s endorsement for our proposal to include speech synthesis in every Macintosh computer and speech recognition in all new types of Macs.

  My wife’s labor continued, and I kept checking the clock. I desperately hoped that she would have the baby in time for me to be there for the birth and also make it back to headquarters in time for the meeting. As I paced the room, my coworkers called and asked if we should cancel the meeting or perhaps have my lieutenant give the presentation to Sculley.

  “No,” I told them. “I think I can make it.”

  But as the labor dragged on, it was looking increasingly unlikely that this would happen, and I was genuinely torn about what I should do: stay by my wife’s side or rush off to an important meeting. Presented with a “problem” like this, my well-trained engineering mind kicked into high gear. I weighed all options in terms of inputs and outputs, maximizing my impact on measurable results.

  Witnessing the birth of my first child would be great, but my daughter would be born whether I was there or not. On the other hand, if I missed this presentation to Sculley, it could have a very substantial and quantifiable impact. Maybe the software wouldn’t respond well to my replacement’s voice—I had a knack for coaxing the best performance out of it—and Sculley might shelve speech-recognition research indefinitely. Or maybe he would greenlight the project but then place someone else in charge of it. I imagined that the fate of artificial intelligence research hung in the balance, and maximizing the chances of success simply meant I had to be in that room for the presentation.

  I was in the midst of these mental calculations when the doctor informed me that they would be performing an immediate cesarean section. My wife was rushed off to an operating room with me in tow, and within an hour Shen-Ling and I were holding our baby daughter. We all had some time together, and with little time left to spare, I took off for the presentation.

  It went extremely well. Sculley both greenlighted the project and demanded a full-on publicity campaign around what I had created. That campaign led to a high-profile TED talk, write-ups in the Wall Street Journal, and an appearance on Good Morning America in 1992, with John Sculley and I demonstrating the technology for millions of viewers. On the program, we used voice commands to schedule an appointment, write a check, and program a VCR, showcasing the earliest examples of futuristic functions that wouldn’t go mainstream for another twenty years, with Apple’s Siri and Amazon’s Alexa. These triumphs filled me with great personal pride and also turbocharged my career.

  But looking back, it’s not those career successes that stick in my mind. It’s the scene in that hospital room. If I had been forced to choose between the birth of my first child and that Apple meeting, I likely would have chosen the meeting.

  Today, I must confess that I find this deeply embarrassing but not entirely baffling. That’s because this wasn’t just about one meeting. It was a manifestation of the machine-like me
ntality that had dominated my life for decades.

  THE IRONMAN

  As a young man, computer science and artificial intelligence resonated with me because the crystal logic of the algorithms mirrored my own way of thinking. At the time, I processed everything in my life—friendships, work, and family time—as variables or inputs in my own mental algorithm. They were things to be quantified and metered out in the precise amounts required to achieve a specific outcome.

  Like any good algorithm, I of course had to balance multiple goals. Self-driving cars don’t just optimize for getting you home as fast as possible; they must do so without breaking any laws and while minimizing the risk of accidents. Likewise, I had to make certain tradeoffs between my personal and professional lives. I hadn’t been a completely absent father, neglectful husband (the episode of my daughter’s birth notwithstanding), or ungrateful son. My social algorithms were good enough that I made a point of remembering anniversaries, giving thoughtful gifts, and spending some time with the people in my family.

  But I approached these as minimization functions, looking for ways to achieve the desired result while putting in the least amount of time possible. I always weighted the master algorithm heavily in favor of my own career goals to maximize time at work, personal influence, and status within my profession.

  When I was given vacations of four weeks, I would spend one or two weeks with my mother in Taiwan or with my family in Beijing and then head right back to work. Even when a surgical procedure forced me to remain lying flat in bed for two weeks, I couldn’t let my work go. I had a metal crane built that suspended a computer monitor above my pillow and connected it with a keyboard and mouse that I could lay across my lap. I was back to answering emails within hours of the surgery.

  I wanted my employees, bosses, and fans to see me as a supercharged productivity machine, someone who did twice the work and needed half the rest of a normal human being. It also gave my team the not-so-subtle suggestion that I expected similar effort from them. My coworkers started calling me by the nickname “Ironman,” and I loved it.

  That work ethic powered an exhilarating lifestyle. I had a chance to stand at the frontier of science, the peak of global business, and in the limelight of national celebrity. In 2013, I was honored as one of the Time 100, the magazine’s list of the most influential people in the world.

  WHAT DO YOU WANT ON YOUR TOMBSTONE?

  Each of those achievements just added more fuel to my internal fire. They pushed me to work harder and to preach this lifestyle to millions of young Chinese people. I wrote best-selling books with titles like Be Your Personal Best and Making a World of Difference. I traveled to college campuses around the country to deliver inspirational speeches. China was reemerging as a global power after centuries of poverty, and I exhorted Chinese students to seize the moment and make their own mark on history.

  Ironically, I concluded these lectures with a striking image: a picture of my own tombstone. I told them that the best way to find one’s calling was to picture your own grave and imagine what you want written on it. I said that my mission was clear, and my tombstone was ready:

  Here lies Kai-Fu Lee,

  scientist and business executive.

  Through his work at top technology companies

  he turned complex technical advances into products

  that everyone could use

  and everyone could benefit from.

  It made for a fantastic conclusion to the speeches, a call to action that resonated with the ambition pulsing through the country at the time. China was evolving and growing as fast as any country in history, and the excitement was palpable. I felt perfectly in my element and at the height of my powers.

  After leaving Google and founding Sinovation Ventures, I began to spend more time mentoring young people. I used my massive following on the Twitter-like platform Weibo to engage directly with Chinese students, offering them guidance and writing open letters that were collected into books. Although I remained the head of one of the country’s most prestigious venture-capital funds, students began referring to me as “Teacher Kai-Fu,” an honorific that in China combined great respect and also a certain closeness.

  I basked in this role as a mentor to millions of students. I believed that this turn toward “teaching” proved my own selflessness and genuine desire to help others. In my speeches at Chinese universities, I kept the tombstone portion but changed the epitaph:

  Here lies Kai-Fu Lee,

  who had a love for education

  during the time of China’s rise.

  Through writing, the internet, and lectures,

  he helped many young students,

  who lovingly called him “Teacher Kai-Fu.”

  Delivering that speech to enraptured audiences gave me a rush. The new epitaph made for an even better ending, I thought, speaking to my substantial influence and also a certain wisdom that came with age. I had gone from scientist to engineer, and from executive to teacher. Along the way, I had managed to maximize my impact on the world while giving my fans a sense of warmth and empathy. The algorithm of my mind, I told myself, had been tuned to perfection.

  It would take an encounter with the reality that lay behind that tombstone—my own mortality—to understand just how foolish and misguided my calculations had been.

  DIAGNOSIS

  The technician in charge of the PET scan was all business. After he showed me into the room, he immediately set about inputting my information and then programming the imaging device. Each year my wife and I traveled back to Taiwan for our medical checkups. Earlier in 2013, one of our close relatives had been diagnosed with cancer, and so my wife decided that this year we would both get MRI and CT scans. After our checkup, my doctor said that he’d found something during the preliminary scans, and that I should come back in for a PET scan.

  While MRI and CT scans require an expert eye to decipher, the results of a PET scan are relatively easy for anyone to understand. Patients are injected with a radioactive tracer, a dose of glucose that contains a tiny amount of a radioisotope. Cancerous cells tend to absorb sugar more intensely than other parts of the body, so these radioisotopes will tend to cluster around potentially cancerous growths. Computer images generated by the scans represent those clusters in bright red. Before we began, I asked the technician if I could see the scan once I was finished.

  “I’m not a radiologist,” he said. “But yes, I can show you the pictures.”

  With that, I lay down on the machine and disappeared into the circular tube within. When I emerged forty-five minutes later, the technician was still hunched over his computer, staring intently at the screen and clicking his mouse in rapid succession.

  “Can I see the pictures now?” I asked.

  “You really should go see your radiologist first,” he replied without looking up.

  “But you told me that I could see it,” I protested. “It’s right on the screen there, isn’t it?”

  Giving in to my insistence, he pivoted the computer monitor around to face me. A cold chill seized my chest, turning into an icy shiver as it spread across my skin. The black scan of my body was dotted with numerous red blotches across my stomach and abdomen.

  “What are all these red things?” I said, my jaw beginning to quiver.

  The technician wouldn’t look me in the eye. I felt that initial chill turning into a hot panic.

  “Are these tumors?” I demanded.

  “There’s a probability that these are tumors,” he replied, still not making eye contact. “But you should really stay calm and go see your radiologist.”

  My mind was swimming, but my body continued on autopilot. I asked the technician to please print the scan for me, and I headed down the hall to the radiologist’s office. I didn’t have an appointment with the radiologist yet, and it was against the rules for them to examine my printouts casually, but I begged and pleaded until someone there agreed to make an exception. After looking over the scans, the radiolog
ist told me that the pattern of these clusters meant that I had lymphoma. When I asked what stage it was in, he tried to deflect the question.

  “Well, it’s complex. We have to find out what kind—”

  I cut him off: “But what stage is it?”

  “Probably stage four.”

  I walked out of the room and then the hospital clutching the paper with both hands, holding it close to my chest so no one passing by could glimpse what was growing inside me. I decided I had to go home and write my will.

  THE WILL

  That teardrop on the page was going to cost me an hour of hard work. I had tried to dab it away with tissue as it grew heavy on my eyelash, but I was a second too late and it dropped to the paper below, landing squarely atop the Chinese character for “Lee.” As the salty tear mixed with the ink on the page, it formed a tiny black puddle that slowly seeped into the paper. I had to start over.

  For a will to be in effect immediately in Taiwan, it must be handwritten, with no blemishes or corrections. It’s a straightforward requirement, if a bit dated. To accomplish this, I took out my best ink pen, the same one I’d used to sign hundreds of copies of the books I had written: a best-selling autobiography and several volumes encouraging young Chinese people to take control of their careers through hard work. That pen was failing me now. My hand quivered with anxiety, and my mind couldn’t shake the image of that PET scan. I tried to remain focused on the lawyer’s instructions for the will, but as my mind wandered, my pen would slip, marring one Chinese character and forcing me to start from scratch.