AI扩散报告：AI最被应用、开发与构建的地方

Executive Summary The Three Forces of Diffusion The history of human progress is, in many ways,the history of technology's diffusion, Among theseadvances, a fewwhat economists call general-purposetechnologies—have had the greatest impact on humanityThe printing press democratized knowledge. The steamengine powered indlustry. Electricity transformed dailylife and enabled computing. And the internet connectedthe world, accelerating the exchange of ideas, trade, andcollaboration across borders. Every transformative technology advance throughthree forces: + Frontier builders — the inventors and pioneers whopush the boundaries of what is possible.In Al, these are the researchers and model creatorsexpanding the limits of intelligence. • Infrastructure builders — the engineers,entrepreneurs, and institutions who scalebreakthroughs through networks, tools, and skillsIn Al, infrostructure burilders provide the compute andaisod aabia a ao o a Each transformed society not through invention alone, butthrough diffusion—the way millions of people made thesetechnologies part of how they live, work, and learn. Artificial intelligence is the next great general-purposetechnology—and the fastest-spreading technology inhuman history. In less than three years, more than 1.2billion people have used Al tools, a rate of adoptionfaster than the internet, the personal computer, or eventhe smartphone. Yet, similar to other general-purposetechnologies before, its benefits are not spreading evenly.Al use in the Global North is roughly double that in theGlobal South. Without focused effort, this gap will definewho benetits trom Al tor decades to come. • Users — the individuals, companies, and governmentswho use, adapt and apply new technologies to solvereal-world problems. History teaches that progress accelerates when all threeevolve together. Edison built the light bulb—but it tookpower grids and everyday users to make electricityuniversal. The same holds true for Al. Whatthe Data Shows Measuring Al's Global Progress As a general-purpose technology, Al stands on theshoulders of three others—electricity, connectivity.and computing. Its adoption is fastest where thesefoundations exist, ancl slowest where they dlo not. Nearlyfour billion peoplehalf the world—still lack the basicsneeded to use Al. To understand this transformation, we propose threeconplermentary indices: • Al Frontier Index — measuring the world's leadingfrontier models by their performance and innovation. • Al Infrastructure Index — capturing where thecapacity to build, train, and scale Al exists. Al adoption in the Global North is roughly twice that of theGlobal South, with the gap widening sharply in countrieswhere GDP per capita falls below $20,000. In someGlobal North countries, more than half of the working-age population uses Al, while in parts of Sub-SaharanAfrica and Asia, some of the least developed nations haveadoption rates below 10%. • Al Diffusion Index — reflecting where Al is beingadopted. Together, they show not only who is building Al, but whocan benefit from it A new barrier is also emerging: language. Nations wherelow-resource languages dominate—like Malawi or Laos—show lower acloption even after adjusting for GDP andinternet access. Ultimately, the value of artificial intelligence will be judgednot by the number of models produced, but by the extentto which they benefit society Some countries—Singapore, the UAE, Norway, andIreland—stand out as leaclers in Al Adoption, provingAoiod pue uogeenpa Kboouupar or ssae uos ecoordination can drive rapid adloption even withoutfrontier-level model development or data centers. From a frontier builder perspective, the number of Al modelscontinues to rise, while the performance gap between themkeeps narrowing. The U.S., led by OpenAl's GPT-5, remainsat the frontier, with China trailing by less than six months.Only seven countries—the U.S, China, France, South Korea,the u.K, Canada, and Israelrank armong the top 200models, and the distance between the frontier (U.S.) and thelast of these (lsrael) is now just 11 months. From an infrastructure builder perspective, the U.S. andChina together host 86% of global data center capacity.underscoring how concentrated the foundation otAl remains. The Importance of Diffusion Thirty years ago, the World Bank published The East AsianMiracle, [1] a landmark report that sought to explain how ahandful of econormies in East Asia achieved unprecedentedgrowth. Its conclusion was clear: adoption and adaption oftechnologies developed elsewhere—rather than inventionof these technologies-—can drive national transformation.Few comparisons illustrate this more clearly than SouthKorea and the Philippines. tax incentives, and infrastructure investnent—Koreansemiconductor manufacturers quickly establishedthemselves as global leaders South Korea's economy surged, growing 6.2 percent annuallyand doubling living standarcs every eleven years. By contrastthe Philippine eco