前沿者的AI扩展指南：来自行业领导者的经验教训

AuthorsSenthil RamaniGlobal Lead forData & AI,AccentureThe front-runners’ guide to scaling AI:Lessons from industry leaders Lan GuanChief AI Officer,Accenture Philippe RoussiereGlobal Lead forInnovation and AI,Accenture Research About the researchWe surveyed 2,000 C-suite and data-scienceexecutives, who lead 1,998 of the world’s largestcompanies (revenues greater than $1 billion), which areheadquartered in 15 countries (Australia, Brazil, Canada,China, Germany, France, India, Italy, Japan, Saudi Arabia,Singapore, Spain, United Arab Emirates, United Kingdomand United States) and operate in nine industries(banking, insurance, energy, consumer goods andservices, life sciences, utilities, retail, public services andcommunications and media). The survey, fielded fromJune to July 2024, aimed to shed light on how companiesdevelop and deploy AI models to create financial andnon-financial value. The survey covered topics suchas organizations’ data and AI strategy, data and AIarchitecture, budgets for—and investments in—strategicbets, talent strategy, ecosystem strategy, responsible AI,AI-related challenges and AI adoption rates.The front-runners’ guide to scaling AI:Lessons from industry leaders To identify the most important strategic bets (see “Getstrategic,” below), we also interviewed numerous C-suiteexperts within and outside Accenture. In addition, wedeployed machine learning to identify both the keycapabilities associated with scaling strategic bets andcompanies’ progress in developing those capabilities.The research was further enriched with insights from ourextensive experience helping clients scale AI solutions.By drawing on these diverse inputs, our findings thuscapture both strategic perspectives on AI and real-worldexecution challenges.For the purposes of this report, “scaling AI” refers tothe process of expanding AI implementation across anenterprise to achieve broader, more impactful outcomes.Scaling includes integrating AI into diverse businessprocesses and workflows; ensuring widespread adoptionacross assets and employees; seamlessly integratingAI with existing systems; driving innovation to gain acompetitive edge in the market; and otherwise improvingkey performance metrics. “Generative AI” is used asan umbrella term for artificial intelligence that canproducebrand-new output—such as text, images, videos,audio and code. Executive summaryFor businesses, securing a sustained advantage over competitorswas long the Holy Grail—a coveted, yet elusive prize. Today,however, generative artificial intelligence and other forms of AIhave flipped the script, bringing the previously unattainable withinreach. That’s why the world’s largest companies are investingheavily in data and AI.But reinventing the enterprise with generative AI (gen AI) isn’tsimply a matter of deploying a few chatbots. Reinvention is aboutbuilding advanced AI capabilities like “agentic architecture,”networks of AI agents that go beyond automating routine tasks toorchestrating entire business workflows.Endowed with sophisticated reasoning, AI agents collaborateautonomously to improve quality, productivity and cost-efficiencyat scale. Agentic architecture is spreading fast: one-third of thecompanies we surveyed for this report are already using AI agentsto strengthen their innovation capabilities.Reinvention thus requires integrating AI deeply into the core ofa company’s strategy. To do this, businesses, under the proactiveleadership of their CEO and board, must go beyond surface-levelapplications of AI and prioritize structural and strategic changesthat unlock AI’s full potential.The front-runners’ guide to scaling AI:Lessons from industry leaders Though every business may want an AI-powered edge, manycompanies are still struggling to advance beyond their initialAI experiments. A big reason for this, our research also shows,is low data “readiness”—which arises when all types of data,especially unstructured data, are not used to the max.Encouragingly, most business leaders recognize this challenge.For example, 70% of the companies we surveyed acknowledgedthe need for a strong data foundation when trying to scale AI.Data, of course, isn’t the only obstacle to enterprise reinventionwith gen AI. Outdated IT systems, as well as workers’ lack of accessto, respectively, gen AI tools, comprehensive training and clearguidance from leadership are significant barriers, too.At the same time, our research revealed that a small minority ofcompanies (“front-runners”) are already achieving considerablesuccess at reinventing their enterprises with gen AI. Thesecompanies consistently get one very important thing right: Theycombine what we call “table stakes” investments in gen AI with“strategic bets” (see below, “Get strategic”).Front-runners, for example, use agentic AI in their table stakes toboost efficiency. And in their strategic bets, they deploy agentic AIto radically reinvent their organizational processes and workflows. 70% of the companieswe