Competing in the Age of AI PDF 2020 Harvard Business Review

t os rP REPRINT R2001C PUBLISH...

t os rP REPRINT R2001C PUBLISHED IN HBR JANUARY–FEBRUARY 2020 ARTICLE yo op TECHNOLOGY Competing in the Age of AI tC How machine intelligence changes the rules of business by Marco Iansiti and Karim R. Lakhani No Do This document is authorized for educator review use only by Taeha Kim, Chung-Ang University until Sep 2020. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  t os rP yo op tC No Do This document is authorized for educator review use only by Taeha Kim, Chung-Ang University until Sep 2020. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  FOR ARTICLE REPRINTS CALL 800-988-0886 OR 617-783-7500, OR VISIT HBR.ORG t Marco Iansiti Karim R. os I L LU ST R ATO R AU T H O RS Professor, Harvard Lakhani Business School Professor, Harvard Business School EDDIE GUY T E C H N O LO GY rP COMPETING AGE in THE yo of AI op tC No How machine intelligence changes the rules of business Do Harvard Business Review January–February 2020 3 This document is authorized for educator review use only by Taeha Kim, Chung-Ang University until Sep 2020. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  t os T E C H N O LO GY rP I N 2 0 1 9, J U S T five years after the Ant Financial Services Group was launched, the number of consumers using its ser- vices passed the one billion mark. Spun out of Alibaba, Ant Financial uses artifi- cial intelligence and data from Alipay—its core mobile-payments platform—to run yo an extraordinary variety of businesses, including consumer lending, money market funds, wealth management, health insurance, credit-rating services, and even an online game that encourages people to reduce their carbon footprint. The company serves more than 10 times as many customers as the largest U.S. banks—with less than one-tenth the number of employees. At its last round of funding, in 2018, it had a valuation of $150 billion—almost half that of JPMorgan op Chase, the world’s most valuable financial-services company. Unlike traditional banks, investment institutions, and insurance companies, Ant Financial is built on a digital core. There are no workers in its “critical path” of operating activities. AI runs the show. There is no manager approving loans, no employee providing financial advice, no represen- tC tative authorizing consumer medical expenses. And without the operating constraints that limit traditional firms, Ant Financial can compete in unprecedented ways and achieve unbridled growth and impact across a variety of industries. The age of AI is being ushered in by the emergence of this The elimination of traditional constraints transforms the new kind of firm. Ant Financial’s cohort includes giants like rules of competition. As digital networks and algorithms are Google, Facebook, Alibaba, and Tencent, and many smaller, woven into the fabric of firms, industries begin to function dif- rapidly growing firms, from Zebra Medical Vision and Way- ferently and the lines between them blur. The changes extend No fair to Indigo Ag and Ocado. Every time we use a service from well beyond born-digital firms, as more-traditional organiza- one of those companies, the same remarkable thing hap- tions, confronted by new rivals, move toward AI-based models pens: Rather than relying on traditional business processes too. Walmart, Fidelity, Honeywell, and Comcast are now tap- operated by workers, managers, process engineers, supervi- ping extensively into data, algorithms, and digital networks sors, or customer service representatives, the value we get to compete convincingly in this new era. Whether you’re is served up by algorithms. Microsoft’s CEO, Satya Nadella, leading a digital start-up or working to revamp a traditional refers to AI as the new “runtime” of the firm. True, managers enterprise, it’s essential to understand the revolutionary and engineers design the AI and the software that makes the impact AI has on operations, strategy, and competition. algorithms work, but after that, the system delivers value Do on its own, through digital automation or by leveraging an ecosystem of providers outside the firm. AI sets the prices on THE AI FACTORY Amazon, recommends songs on Spotify, matches buyers and At the core of the new firm is a decision factory—what we call sellers on Indigo’s marketplace, and qualifies borrowers for the “AI factory.” Its software runs the millions of daily ad auc- an Ant Financial loan. tions at Google and Baidu. Its algorithms decide which cars 4 Harvard Business Review January–February 2020 This document is authorized for educator review use only by Taeha Kim, Chung-Ang University until Sep 2020. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  FOR ARTICLE REPRINTS CALL 800-988-0886 OR 617-783-7500, OR VISIT HBR.ORG t os The AI that drives the explosive growth of a firm often isn’t even all that sophisticated. AI doesn’t need to be the stuff of science fiction, simulating human reasoning. rP offer rides on Didi, Grab, Lyft, and Uber. It sets the prices of that their suggestions are having the intended effect. The headphones and polo shirts on Amazon and runs the robots fourth is infrastructure, the systems that embed this process that clean floors in some Walmart locations. It enables cus- in software and connect it to internal and external users. tomer service bots at Fidelity and interprets X-rays at Zebra Take a search engine like Google or Bing. As soon as Medical. In each case the AI factory treats decision-making someone starts to type a few letters into the search box, as a science. Analytics systematically convert internal and algorithms dynamically predict the full search term on the external data into predictions, insights, and choices, which in basis of terms that many users have typed in before and this yo turn guide and automate operational workflows. particular user’s past actions. These predictions are captured Oddly enough, the AI that can drive the explosive growth in a drop-down menu (the “autosuggest box”) that helps the of a digital firm often isn’t even all that sophisticated. To user zero in quickly on a relevant search. Every keystroke bring about dramatic change, AI doesn’t need to be the stuff and every click are captured as data points, and every data of science fiction—indistinguishable from human behavior point improves the predictions for future searches. AI also or simulating human reasoning, a capability sometimes generates the organic search results, which are drawn from referred to as “strong AI.” You need only a computer system a previously assembled index of the web and optimized to be able to perform tasks traditionally handled by people— according to the clicks generated on the results of previous op what is often referred to as “weak AI.” searches. The entry of the term also sets off an automated With weak AI, the AI factory can already take on a range of auction for the ads most relevant to the user’s search, the critical decisions. In some cases it might manage information results of which are shaped by additional experimentation businesses (such as Google and Facebook). In other cases and learning loops. Any click on or away from the search it will guide how the company builds, delivers, or operates query or search results page provides useful data. The more actual physical products (like Amazon’s warehouse robots searches, the better the predictions, and the better the tC or Waymo, Google’s self-driving car service). But in all cases predictions, the more the search engine is used. digital decision factories handle some of the most critical processes and operating decisions. Software makes up the core of the firm, while humans are moved to the edge. REMOVING LIMITS TO SCALE, SCOPE, AND LEARNING COPYRIGHT © 2019 HARVARD BUSINESS SCHOOL PUBLISHING CORPORATION. ALL RIGHTS RESERVED. Four components are essential to every factory. The first The concept of scale has been central in business since at is the data pipeline, the semiautomated process that gathers, least the Industrial Revolution. The great Alfred Chandler cleans, integrates, and safeguards data in a systematic, sus- described how modern industrial firms could reach unprec- tainable, and scalable way. The second is algorithms, which edented levels of production at much lower unit cost, giving No generate predictions about future states or actions of the large firms an important edge over smaller rivals. He also business. The third is an experimentation platform, on which highlighted the benefits companies could reap from the abil- hypotheses regarding new algorithms are tested to ensure ity to achieve greater production scope, or variety. The push IDEA IN BRIEF THE MARKET CHANGE THE CHALLENGE THE UPSHOT Do We’re seeing the emergence of a The AI-driven operating model For digital start-ups and traditional firms new kind of firm—one in which is blurring the lines that used to alike, it’s essential to understand the artificial intelligence is the main separate industries and is upending revolutionary impact AI has on operations, source of value creation and delivery. the rules of business competition. strategy, and competition. Harvard Business Review January–February 2020 5 This document is authorized for educator review use only by Taeha Kim, Chung-Ang University until Sep 2020. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  t os MICROSOFT’S AI TRANSFORMATION rP Microsoft’s trans- with product experience to run Today Core Engineering— component library, algorithm formation into an IT and build the “AI factory” as the IT operation is now repository, and data catalog, AI-driven firm took that would be the foundation known—is a showcase for all used to rapidly enable years of research but gained of the firm’s new operating Microsoft’s own transforma- and deploy digital processes steam with the reorganization model. “Our product is the tion. Thanks to the group’s across different lines of of its internal IT and data process,” DelBene told us. work, many traditional business. assets, which had been dis- “First, we are going to articu- processes that used to be per- Beyond increasing produc- persed across the company’s late what the vision should be formed in silos are enabled by tivity and scalability, the AI various operations. That effort for the systems and processes one consistent software base also helps head off problems. yo was led by Kurt DelBene, the we support. Second, we’re residing in Microsoft’s Azure “We leverage AI to know when former head of Microsoft’s going to be run like a product cloud. In addition, the team things are starting to behave Office business, who’d left to development team. And we’re is driving toward a common in unexpected ways,” DelBene help fix the U.S. government’s going to be agile-based.” To data architecture across the says. “The best we could do HealthCare.gov site before strengthen that orientation on company. The new, AI-based in the past is react as fast as returning to Microsoft in 2015. his team, he brought in hand- operating platform connects possible. Now we can preempt There’s a reason that CEO picked leaders and engineers the sprawling organization things, from bad contracts to Satya Nadella chose someone from the product functions. with a shared software- cyberbreaches.” op for improvement and innovation added a third requirement We call this kind of confrontation a “collision.” As both for firms: learning. Scale, scope, and learning have come learning and network effects amplify volume’s impact on to be considered the essential drivers of a firm’s operating value creation, firms built on a digital core can overwhelm performance. And for a long time they’ve been enabled by traditional organizations. Consider the outcome when tC carefully defined business processes that rely on labor and Amazon collides with traditional retailers, Ant Financial management to deliver products and services to customers— with traditional banks, and Didi and Uber with traditional and that are reinforced by traditional IT systems. taxi services. As Clayton Christensen, Michael Raynor, and After hundreds of years of incremental improvements Rory McDonald argued in “What Is Disruptive Innovation?” to the industrial model, the digital firm is now radically (HBR, December 2015), such competitive upsets don’t fit the changing the scale, scope, and learning paradigm. AI-driven disruption model. Collisions are not caused by a particular processes can be scaled up much more rapidly than tradi- innovation in a technology or a business model. They’re tional processes can, allow for much greater scope because the result of the emergence of a completely different kind No they can easily be connected with other digitized businesses, of firm. And they can fundamentally alter industries and and create incredibly powerful opportunities for learning and reshape the nature of competitive advantage. improvement—like the ability to produce ever more accurate Note that it can take quite a while for AI-driven operat- and sophisticated customer-behavior models and then tailor ing models to generate economic value anywhere near the services accordingly. value that traditional operating models generate at scale. In traditional operating models, scale inevitably reaches Network effects produce little value before they reach a point at which it delivers diminishing returns. But we critical mass, and most newly applied algorithms suffer don’t necessarily see this with AI-driven models, in which from a “cold start” before acquiring adequate data. Ant the return on scale can continue to climb to previously Financial grew rapidly, but its core payment service, Do unheard-of levels. (See the exhibit “How AI-Driven Com- Alipay, which had been launched in 2004 by Alibaba, took panies Can Outstrip Traditional Firms.”) Now imagine what years to reach its current volume. This explains why exec- happens when an AI-driven firm competes with a traditional utives ensconced in the traditional model have a difficult firm by serving the same customers with a similar (or better) time at first believing that the digital model will ever catch value proposition and a much more scalable operating model. up. But once the digital operating model really gets going, 6 Harvard Business Review January–February 2020 This document is authorized for educator review use only by Taeha Kim, Chung-Ang University until Sep 2020. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  FOR ARTICLE REPRINTS CALL 800-988-0886 OR 617-783-7500, OR VISIT HBR.ORG t it can deliver far superior value and quickly overtake traditional firms. os Collisions between AI-driven and traditional firms are happening across industries: software, financial services, retail, telecommunications, media, health care, automobiles, and even agribusiness. It’s hard to think of a business that T EC H N O LO GY isn’t facing the pressing need to digitize its operating model rP and respond to the new threats. such transformations, see the sidebar “Putting AI at the Firm’s Core.”) REBUILDING TRADITIONAL ENTERPRISES Fidelity Investments is using AI to enable processes in For leaders of traditional firms, competing with digital important areas, including customer service, customer rivals involves more than deploying enterprise software or insights, and investment recommendations. Its AI initiatives even building data pipelines, understanding algorithms, build on a multiyear effort to integrate data assets into one and experimenting. It requires rearchitecting the firm’s digital core and redesign the organization around it. The yo organization and operating model. For a very, very long work is by no means finished, but the impact of AI is already time, companies have optimized their scale, scope, and evident in many high-value use cases across the company. learning through greater focus and specialization, which led To take on Amazon, Walmart is rebuilding its operating to the siloed structures that the vast majority of enterprises model around AI and replacing traditional siloed enterprise today have. Generations of information technology didn’t software systems with an integrated, cloud-based architec- change this pattern. For decades, IT was used to enhance the ture. That will allow Walmart to use its unique data assets performance of specific functions and organizational units. in a variety of powerful new applications and automate or Traditional enterprise systems often even reinforced silos enhance a growing number of operating tasks with AI and op and the divisions across functions and products. analytics. At Microsoft, Nadella is betting the company’s Silos, however, are the enemy of AI-powered growth. future on a wholesale transformation of its operating model. Indeed, businesses like Google Ads and Ant Financial’s (See the sidebar “Microsoft’s AI Transformation.”) MyBank deliberately forgo them and are designed to leverage an integrated core of data and a unified, consistent code base. When each silo in a firm has its own data and code, RETHINKING STRATEGY AND CAPABILITIES tC internal development is fragmented, and it’s nearly impos- As AI-powered firms collide with traditional businesses, sible to build connections across the silos or with external competitive advantage is increasingly defined by the ability business networks or ecosystems. It’s also nearly impossible to shape and control digital networks. (See “Why Some to develop a 360-degree understanding of the customer that Platforms Thrive and Others Don’t,” HBR, January–February both serves and draws from every department and function. 2019.) Organizations that excel at connecting businesses, So when firms set up a new digital core, they should avoid aggregating the data that flows among them, and extracting creating deep organizational divisions within it. its value through analytics and AI will have the upper hand. While the transition to an AI-driven model is challeng- Traditional network effects and AI-driven learning curves No ing, many traditional firms—some of which we’ve worked will reinforce each other, multiplying each other’s impact. with—have begun to make the shift. In fact, in a recent You can see this dynamic in companies such as Google, Face- study we looked at more than 350 traditional enterprises in book, Tencent, and Alibaba, which have become powerful both service and manufacturing sectors and found that the “hub” firms by accumulating data through their many net- majority had started building a greater focus on data and work connections and building the algorithms necessary to analytics into their organizations. Many—including Nord- heighten competitive advantages across disparate industries. strom, Vodafone, Comcast, and Visa—had already made Meanwhile, conventional approaches to strategy that important inroads, digitizing and redesigning key compo- focus on traditional industry analysis are becoming increas- nents of their operating models and developing sophisti- ingly ineffective. Take automotive companies. They’re Do cated data platforms and AI capabilities. You don’t have to facing a variety of new digital threats, from Uber to Waymo, be a software start-up to digitize critical elements of your each coming from outside traditional industry boundaries. business—but you do have to confront silos and fragmented But if auto executives think of cars beyond their traditional legacy systems, add capabilities, and retool your culture. industry context, as a highly connected, AI-enabled service, (For a closer look at the key principles that should drive they can not only defend themselves but also unleash new Harvard Business Review January–February 2020 7 This document is authorized for educator review use only by Taeha Kim, Chung-Ang University until Sep 2020. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  t opportunities, it seems inevitable that they will also cause widespread dislocation in many occupations. os The dislocations will include not only job replacement but also the erosion of traditional capabilities. In almost every setting, AI-powered firms are taking on highly specialized T E C H N O LO GY organizations. In an AI-driven world, the requirements for competition have less to do with specialization and more to rP do with a universal set of capabilities in data sourcing, pro- value—through local commerce opportunities, ads, news cessing, analytics, and algorithm development. These new and entertainment feeds, location-based services, and so on. universal capabilities are reshaping strategy, business design, The advice to executives was once to stick with businesses and even leadership. Strategies in very diverse digital and they knew, in industries they understood. But synergies in networked businesses now look similar, as do the drivers of algorithms and data flows do not respect industry boundar- operating performance. Industry expertise has become less ies. And organizations that can’t leverage customers and data critical. When Uber looked for a new CEO, the board hired across those boundaries are likely to be at a big disadvantage. someone who had previously run a digital firm—Expedia— yo Instead of focusing on industry analysis and on the man- not a limousine services company. agement of companies’ internal resources, strategy needs to We’re moving from an era of core competencies that focus on the connections firms create across industries and differ from industry to industry to an age shaped by data and the flow of data through the networks the firms use. analytics and powered by algorithms—all hosted in the cloud All this has major implications for organizations and their for anyone to use. This is why Alibaba and Amazon are able to employees. Machine learning will transform the nature of compete in industries as disparate as retail and financial ser- almost every job, regardless of occupation, income level, vices, and health care and credit scoring. These sectors now or specialization. Undoubtedly, AI-based operating models have many similar technological foundations and employ op can exact a real human toll. Several studies suggest that common methods and tools. Strategies are shifting away from perhaps half of current work activities may be replaced traditional differentiation based on cost, quality, and brand by AI-enabled systems. We shouldn’t be too surprised by equity and specialized, vertical expertise and toward advan- that. After all, operating models have long been designed to tages like business network position, the accumulation of make many tasks predictable and repeatable. Processes for unique data, and the deployment of sophisticated analytics. scanning products at checkout, making lattes, and removing tC hernias, for instance, benefit from standardization and don’t require too much human creativity. While AI improvements THE LEADERSHIP CHALLENGE will enrich many jobs and generate a variety of interesting Though it can unleash enormous growth, the removal of operating constraints isn’t always a good thing. Frictionless systems are prone to instability and hard to stop once they’re in motion. Think of a car without brakes or a skier who can’t How AI-Driven Companies Can slow down. A digital signal—a viral meme, for instance—can spread rapidly through networks and can be just about Outstrip Traditional Firms No impossible to halt, even for the organization that launched it in the first place or an entity that controls the key hubs The value that in a network. Without friction, a video inciting violence or scale delivers eventually tapers a phony or manipulative headline can quickly spread to off in traditional Traditional billions of people on a variety of networks, even morphing to operating models, operating model optimize click-throughs and downloads. If you have a mes- but in digital operating models, sage to send, AI offers a fantastic way to reach vast numbers it can climb much of people and personalize that message for them. But the higher. marketer’s paradise can be a citizen’s nightmare. Do Digital operating models can aggregate harm along Digital operating with value. Even when the intent is positive, the potential model downside can be significant. A mistake can expose a large Value digital network to a destructive cyberattack. Algorithms, if Number of users left unchecked, can exacerbate bias and misinformation on 8 Harvard Business Review January–February 2020 This document is authorized for educator review use only by Taeha Kim, Chung-Ang University until Sep 2020. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  FOR ARTICLE REPRINTS CALL 800-988-0886 OR 617-783-7500, OR VISIT HBR.ORG t os PUTTING AI AT THE FIRM’S CORE rP The transition from analytics, and AI requires However, many organizations updates, protecting against a traditional firm some centralization and fail to realize that they need cyberattacks, and running to an AI-driven a lot of consistency. Data to systematically hire a very help desks. Developing organization cannot happen assets should be integrated different kind of talent and operating-model software in a skunkworks or be spear- across a range of applications set up career paths and is a different game. headed by some separate au- to maximize their impact. incentive systems for those tonomous group. It requires a Fragmented data will be employees. Multidisciplinary holistic effort. In our research virtually impossible to governance. The governance and our work with a variety safeguard consistently, An agile “product” focus. of digital assets has become yo of companies, we’ve come especially given privacy and Building an AI-centric increasingly important and up with five principles that security considerations. operating model is about complex and calls for well- should guide transformations If the data isn’t all held in taking traditional processes thought-out collaboration (beyond common best prac- centralized repositories, then and transforming them across disparate disciplines tices for leading change): the organization must at least into software. Developing a and functions. The challenges have an accurate catalog of product-focused mentality is of data privacy, algorithmic One strategy. Rearchitecting where the data is, explicit essential to getting this done. bias, and cybersecurity are a company’s operating model guidelines for what to do with Like the product managers increasing risk and even means rebuilding each busi- it (and how to protect it), and at any world-class software government intervention ness unit on a new, integrated standards for when and how to development project, the IT and regulation. Governance op foundation of data, analytics, store it so that it can be used teams deploying AI-centered should integrate a legal and and software. This challeng- and reused by multiple parties. applications should have a corporate affairs function, ing and time-consuming deep understanding of the which may even be involved undertaking demands focus The right capabilities. use cases they’re enabling— in product and technology and a consistent top-down Though building a base a product management decisions. AI requires deep mandate to coordinate and of software, data science, orientation that goes well thinking about legal and inspire the many bottom-up and advanced analytics beyond the approach of ethical challenges, including efforts involved. capabilities will take time, traditional IT organizations. careful consideration of what much can be done with a In the past, IT was largely data should be stored and tC A clear architecture. A new small number of motivated, about keeping old systems preserved (and what data approach based on data, knowledgeable people. working, deploying software should not). a massive scale. Risks can be greatly magnified. Consider the growth is dangerous. The potential for businesses that way that digital banks are aggregating consumer savings in embrace digital operating models is huge, but the capacity No an unprecedented fashion. Ant Financial, which now oper- to inflict widespread harm needs to be explicitly considered. ates one of the largest money market funds in the world, is Navigating these opportunities and threats will be a real test entrusted with the savings of hundreds of millions of Chinese of leadership for both businesses and public institutions. consumers. The risks that presents are significant, especially HBR Reprint R2001C for a relatively unproven institution. Digital scale, scope, and learning create a slew of new MARCO IANSITI is the David Sarnoff Professor of Business challenges—not just privacy and cybersecurity problems, but Administration at Harvard Business School, where he heads social turbulence resulting from market concentration, dislo- the Technology and Operations Management Unit and the Digital Initiative. He has advised many companies in the technology sector, cations, and increased inequality. The institutions designed Do including Microsoft, Facebook, and Amazon. KARIM R. LAKHANI is to keep an eye on business—regulatory bodies, for example— the Charles Edward Wilson Professor of Business Administration are struggling to keep up with all the rapid change. and the Dorothy and Michael Hintze Fellow at Harvard Business In an AI-driven world, once an offering’s fit with a market School and the founder and codirector of the Laboratory for is ensured, user numbers, engagement, and revenues can Innovation Science at Harvard. They are the coauthors of the book skyrocket. Yet it’s increasingly obvious that unconstrained Competing in the Age of AI (Harvard Business Review Press, 2020). Harvard Business Review January–February 2020 9 This document is authorized for educator review use only by Taeha Kim, Chung-Ang University until Sep 2020. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860