Page 44 - Fortune-November 01, 2018
P. 44

ar tifi c i al  intelli gence  +  kai -fu  lee




         web, Internet A.I. leverages the fact that users  and then using that information to make decisions (apply-
         automatically label data as we browse: buying  ing pressure to the brake in order to slowly stop the vehicle).
         vs. not buying, clicking vs. not clicking. These  In the area of robotics, such advanced A.I. algorithms will be
         cascades of labeled data build a detailed profile  applied to industrial applications (automated assembly lines
         of our personalities, habits, demands, and    and warehouses), commercial tasks (dishwashing and fruit-
         desires: the perfect recipe for more tailored  harvesting robots), and eventually consumer ones too.
         content to keep us on a given platform, or to
         maximize revenue or profit.
           The second wave is “business A.I.” Here,    THE CHANGES YET TO COME
         algorithms can be trained on proprietary data
         sets ranging from customer purchases to ma-
         chine maintenance records to complex business  BECAUSE A.I. CAN BE PROGRAMMED to maximize profitability or
         processes—and ultimately lead managers to     replace human labor, it adds immediate value to the economy.
         improved decision-making. An algorithm, for   A.I. is fast, accurate, works around-the-clock, doesn’t complain,
         example, might study many thousands of bank   and can be applied to many tasks, with substantial economic
         loans and repayment rates, and learn if one   benefit. How substantial? PwC estimates that the technology
         type of borrower is a hidden risk for default  will contribute about $16 trillion to worldwide GDP by 2030.
         or, alternatively, a surprisingly good, but over-  But that gift doesn’t come without challenges to human-
         looked, lending prospect. Medical researchers,  ity. The first and foremost is job displacement: Since A.I. can
         similarly, can use deep-learning algorithms to di-  perform single tasks with superhuman accuracy—and most
         gest enormous quantities of data on patient di-  human jobs are single-task—it follows that many routine jobs
         agnoses, genomic profiles, resultant therapies,  will be replaced by this next-generation tech. That includes both
         and subsequent health outcomes and perhaps    white-collar and blue-collar jobs. A.I. also faces questions with
         discover a worthy personalized treatment proto-  security, privacy, data bias, and monopoly maintenance. All are
         col that would have otherwise been missed. By  significant issues with no known solution, so governments and
         scouting out hidden correlations that escape our  corporations should start working on them now.
         linear cause-and-effect logic, business A.I. can  But one concern we don’t have to face quite yet is the one that
         outperform even the most veteran of experts.  may be most common these days, cast in the image of science-
           The third wave of artificial intelligence—   fiction movies—that machines will achieve true human-level
         call it “perception A.I.”— gets an upgrade with  (or even superhuman-level) intelligence, making them capable
         eyes, ears, and myriad other senses, collecting  presumably of threatening mankind.
         new data that was never before captured, and    We’re nowhere near that. Today’s A.I. isn’t “general artificial
         using it to create new applications. As sensors  intelligence” (the human kind, that is), but rather narrow—lim-
         and smart devices proliferate through our     ited to a single domain. General A.I. requires advanced capabili-
         homes and cities, we are on the verge of enter-  ties like reasoning, conceptual learning, common sense, planning,
         ing a trillion-sensor economy. This includes  cross-domain thinking, creativity, and even self-awareness and
         speech interfaces (from Alexa and Siri to     emotions, which remain beyond our reach. There are no known
         future supersmart assistants that remember    engineering paths to evolve toward the general capabilities above.
         everything for you) as well as computer-vision  How far are we from general A.I.? I don’t think we even
         applications—from face recognition to manu-   know enough to estimate. We would need dozens of big break-
         facturing quality inspection.                 throughs to get there, when the field of A.I. has seen only one
           The fourth wave is the most monumental      true breakthrough in 60 years. That said, narrow A.I. will bring
         but also the most difficult: “autonomous A.I.”  about a technology revolution the magnitude of the Industrial
         Integrating all previous waves, autonomous A.I.  Revolution or larger—and one that’s happening much faster.
         gives machines the ability to sense and respond  It’s incumbent upon us to understand its monumental impact,
         to the world around them, to move intuitively,  widespread benefits, and serious challenges.
         and to manipulate objects as easily as a hu-
         man can. Included in this wave are autono-
         mous vehicles that can “see” the environment  This essay is adapted from Lee’s new book, AI Superpowers:
         around them: recognizing patterns in the      China, Silicon Valley, and the New World Order (Houghton
         camera’s pixels (red octagons, for instance);  Mifflin Harcourt). He is the chairman and CEO of Sinovation
         figuring out what they correlate to (stop signs);  Ventures and the former president of Google China.





         94
         FO R T U N E. CO M //  N O V. 1 . 1 8
   39   40   41   42   43   44   45   46   47   48   49