GTIPA Perspectives: Making AI Policy to Drive Innovation
Artificial intelligence (AI) is the general-purpose technology of this decade, and possibly even this century, with impacts predicted to be as deep and diffuse as electricity and the Internet. As with other general-purpose technologies in their nascent stages, much still remains uncertain regarding AI’s capacity and direction. This report addresses the state of innovation policy on AI around the world, compiling 32 vignettes from 28 think tanks in 30 countries—Albania, Argentina, Australia, Bangladesh, Bosnia and Herzegovina, Brazil, Bulgaria, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Germany, Greece, India, Indonesia, Italy, Kosovo, Mexico, Montenegro, North Macedonia, Pakistan, the Philippines, Poland, Peru, Serbia, South Africa, Sweden, the United States, and Vietnam. It details the impactful measures that these governments have taken to drive successful development and deployment of AI in their economies. The countries highlighted in this report are diverse, representing every continent except Antarctica and a range of economic and social contexts. Some of these vignettes portray triumphs already won, some outline visions yet to be realized. All of them support GTIPA’s belief that, if governed wisely, AI can make substantial contributions to global growth and development.
This report also includes a Declaration of Open AI, which calls for democratic countries to establish a global framework for free and open AI, protecting freedom of expression, innovation, free trade, and access to AI while promoting transparency and responsible use. The declaration, which is endorsed by 23 signatories, argues that governments should avoid censoring or manipulating AI and should instead pursue international collaboration to ensure AI remains accessible, innovative, and aligned with democratic values.
The Promise of AI
While AI has long been of interest to academic researchers, over the last 10 years the technology has evolved from a research frontier to a general-purpose tool reshaping the global economy. In 2016, Google DeepMind's AlphaGo system defeated the 18-time world-champion of Go, a game long considered a benchmark of AI.1 In 2023, OpenAI’s GPT-4 not only passed the Uniform Bar Exam, but also scored in the top tier of human test-takers. The same year, ChatGPT also passed the U.S. Medical Licensing Examination.2 Today, large language models from Anthropic, OpenAI, and other AI leaders can write and debug software, analyze complex legal and financial documents, and execute intricate tasks with little supervision.
The value of such a tool to the economy is expected to be massive, though calculating the exact impact will be the project of future researchers. Yet the predicted estimates are staggering. Goldman Sachs estimates that generative AI could drive a 7 percent increase in global gross domestic product (GDP), corresponding with $7 trillion in value, and increase productivity growth by 1.5 percentage points over 10 years.3 Estimates from the World Economic Forum and EYParthenon predict similar effects to labor productivity.4 Whatever the macroeconomic effects turn out to be, the impacts of AI adoption at the firm and worker level are already becoming clear. For example, economists found that AI-enabled customer support systems produced a 34 percent productivity improvement for novice and low-skilled workers.5 Researchers have also found that use of AI increases the output of software developers by 26 percent.6
In addition to the prosperity promised by labor productivity gains in the private sector, AI has also proven its utility for public sector and social impact projects. It is already being used to provide educational solutions to youth in Iraq, Tanzania, South Korea, and the United States, digitize and translate documents in India, or map forests in Brazil.7 As the case studies in this report show, it has been adopted by governments around the world to provide public services in health care, education, and criminal justice.
Despite its evident promise, the course of AI’s development, deployment, and adoption will be influenced in large part by the decisions of governments around the world. This publication examines the consequential choices about infrastructure, investment, talent, and regulation that policymakers are making today to determine what is working and how we might improve.