{ "title": "The Macroeconomics of Artificial Intelligence", "sourceUrl": "https://www.imf.org/en/publications/fandd/issues/2023/12/macroeconomics-of-artificial-intelligence-brynjolfsson-unger", "canonical": "https://www.imf.org/en/publications/fandd/issues/2023/12/macroeconomics-of-artificial-intelligence-brynjolfsson-unger", "overlayPath": "/en/publications/fandd/issues/2023/12/macroeconomics-of-artificial-intelligence-brynjolfsson-unger/index.md", "summary": "The collective decisions we make today will determine how AI affects productivity growth, income inequality, and industrial concentration", "sections": [ { "heading": "Overview", "content": "- Focus: implications of AI on productivity growth, the labor market, and industrial concentration.\n- Central claim: AI’s economic impact is not predetermined; collective technological and policy choices today will shape divergent futures.\n- Good policy requirements listed in the article:\n - Creative policy experiments\n - A set of positive goals for what society wants from AI, not just negative outcomes to be avoided\n - Understanding that the technological possibilities of AI are deeply uncertain and rapidly evolving and that society must be flexible in evolving with them" }, { "heading": "Productivity growth", "content": "- Context:\n - The US economy has been stuck with disturbingly low productivity growth for most of the past 50 years, except for a brief resurgence in the late 1990s and early 2000s.\n - Productivity—output per unit of input—largely determines the wealth of nations and living standards.\n- Low-productivity future (path of least resistance):\n - AI adoption may remain slow and confined to large firms.\n - AI could be a narrow labor-saving technology (a “so-so technology”) rather than enabling novel or powerful worker complements.\n - Displaced workers might end up in less productive jobs, muting aggregate productivity gains.\n - Economic gains may appear only decades after initial technological promise (Solow paradox).\n - Organizational and managerial failures could blunt AI’s benefits.\n - Legal and regulatory hurdles may arise around intellectual property, potentially creating a “patent thicket” that prevents models from being trained on data without clear rights.\n - National regulators may impose strict regulations or bans that slow development and dissemination.\n- High-productivity future:\n - AI could be applied to a substantial share of tasks done by most workers (Eloundou and others 2023), massively boosting productivity in those tasks.\n - AI might complement workers, freeing time for nonroutine, creative, and inventive tasks.\n - AI could capture tacit knowledge from individuals and organizations using vast newly digitized data.\n - Result: a permanently higher growth rate, broader applicability when integrated with robots, and AI-backed research enabling radical advances in medicine, math, science, and recursive self-improvement." }, { "heading": "Income inequality", "content": "- Context:\n - Increase in income inequality between individual workers over the past 40 years is a major concern.\n - Past computerization and IT automated routine middle-income jobs, polarizing the labor force.\n- Higher-inequality future:\n - AI substitutes directly for many kinds of human labor, driving down wages for many workers.\n - Generative AI produces words, images, and sounds—tasks once considered nonroutine and creative—expanding job threats.\n - Entire industries may be upended; displacement concentrated into low-paying service jobs (hospital orderlies, nannies, doormen).\n - Outcome: greater polarization into a small high-skilled elite and a large underclass of poorly paid service workers.\n - This is not a future of mass unemployment, but of increased inequality.\n- Lower-inequality future:\n - AI assists least experienced or least knowledgeable workers to become more productive (example: software coders using Copilot).\n - Study cited: among 5,000 workers in complex customer assistance jobs, those given an AI assistant had the greatest productivity gains among the least skilled or newest workers (Brynjolfsson, Li, and Raymond 2023).\n - If employers share productivity gains with workers, income distribution could become more equal.\n - AI may remove tedious routine work and complement creative tasks, improving work experience, reducing turnover, and increasing customer satisfaction." }, { "heading": "Industrial concentration", "content": "- Context:\n - Since the early 1980s, industrial concentration has risen dramatically in the United States and many other advanced economies; large “superstar” firms are more capital-intensive and technologically sophisticated.\n- Higher-concentration future:\n - Only the largest firms intensively use AI, becoming more productive, profitable, and larger.\n - AI model development is expensive in raw computational power and massive datasets; examples:\n - GPT-4 initial development cost: more than $100 million to train\n - GPT-4 operating cost: about $700,000 a day to run\n - Typical cost of developing a large AI model may soon be in the billions of dollars\n - Scaling laws relating training costs to improved performance favor firms with the biggest budgets and datasets.\n - Large firms may gain further advantages via AI-enabled internal coordination, strengthening the “visible hand” of top management.\n- Lower-concentration future:\n - Open-source AI models (examples in the article: Meta’s LLaMA, Berkeley’s Koala) become widely available, creating a vibrant open-source ecosystem.\n - Small businesses gain access to industry-leading production technologies they could not otherwise afford.\n - Internal memo (May 2023) cited: “open-source models are faster, more customizable, more private, and pound-for-pound more capable” than proprietary models; small open-source models can be repeated quickly and trained more cheaply.\n - AI could encourage decentralized innovation, altering firm boundaries and enabling more innovators to found small firms." }, { "heading": "Toward a policy agenda and research priorities", "content": "- Core point: the worse futures (low productivity, higher inequality, higher concentration) are the path of least resistance; reaching better futures requires active policy shaping.\n- Policy questions to guide framing:\n - How can policies encourage AI that complements human labor instead of replacing it?\n - What choices will encourage AI development that firms of all sizes can access?\n - What kind of open-source ecosystem is required, and how can policymakers support it?\n - How should AI labs approach model development, and how should firms approach AI implementation?\n - How does society get AI that unleashes radical innovation rather than marginal tweaks?\n- Actors with power to affect direction:\n - Major corporations, AI/computer science university labs, federal legislators and regulators, local policymakers, voters, and labor unions.\n- Research needs:\n - There is a deep imbalance between research advancing AI frontiers and research understanding AI’s economic and social consequences.\n - Given potential effects measured in trillions of dollars, far greater investment should be made in research on the economics of AI.\n - Society needs innovations in economic and policy understanding that match the scale and scope of AI breakthroughs." }, { "heading": "Authors", "content": "- ERIK BRYNJOLFSSON: Jerry Yang and Akiko Yamazaki Professor at the Stanford Institute for Human-Centered AI; directs the Stanford Digital Economy Lab.\n- GABRIEL UNGER: Postdoctoral fellow at the Stanford Digital Economy Lab.\n\nSource: The Macroeconomics of Artificial Intelligence, ERIK BRYNJOLFSSON and GABRIEL UNGER, December 2023.\n\n---\n\n Content in this bundle\n\n- F&D : THE MACROECONOMICS OF Artificial Intelligence\n - F&D : THE MACROECONOMICS OF Artificial Intelligence (Markdown version){rel=\"alternate\" type=\"text/markdown\"}\n - F&D : THE MACROECONOMICS OF Artificial Intelligence (PDF){rel=\"external\" type=\"application/pdf\"}\n\n---\n\nSource: https://www.imf.org/en/publications/fandd/issues/2023/12/macroeconomics-of-artificial-intelligence-brynjolfsson-unger" } ], "bullets": [ "[Markdown version](/en/publications/fandd/issues/2023/12/macroeconomics-of-artificial-intelligence-brynjolfsson-unger/index.md)", "[Structured JSON version](/en/publications/fandd/issues/2023/12/macroeconomics-of-artificial-intelligence-brynjolfsson-unger/index.json)", "[Bundle manifest](/en/publications/fandd/issues/2023/12/macroeconomics-of-artificial-intelligence-brynjolfsson-unger/bundle-manifest.json)", "Authors: Erik Brynjolfsson, GABRIEL UNGER", "Published: December 2, 2023", "Focus: implications of AI on productivity growth, the labor market, and industrial concentration.", "Central claim: AI’s economic impact is not predetermined; collective technological and policy choices today will shape divergent futures.", "Good policy requirements listed in the article:", "Context:", "Low-productivity future (path of least resistance):", "High-productivity future:", "Context:", "Higher-inequality future:", "Lower-inequality future:", "Context:", "Higher-concentration future:", "Lower-concentration future:", "Core point: the worse futures (low productivity, higher inequality, higher concentration) are the path of least resistance; reaching better futures requires active policy shaping.", "Policy questions to guide framing:", "Actors with power to affect direction:", "Research needs:", "ERIK BRYNJOLFSSON: Jerry Yang and Akiko Yamazaki Professor at the Stanford Institute for Human-Centered AI; directs the Stanford Digital Economy Lab.", "GABRIEL UNGER: Postdoctoral fellow at the Stanford Digital Economy Lab.", "**F&D : THE MACROECONOMICS OF Artificial Intelligence**" ], "related": [ { "title": "F&D : THE MACROECONOMICS OF Artificial Intelligence", "role": "document", "sourceUrl": "https://www.imf.org/-/media/files/publications/fandd/article/2023/december/20-25-brynjolfsson-final.pdf", "summary": { "path": "/-/media/files/publications/fandd/article/2023/december/20-25-brynjolfsson-final.pdf.md", "mime": "text/markdown" }, "binary": { "path": "/-/media/files/publications/fandd/article/2023/december/20-25-brynjolfsson-final.pdf", "mime": "application/pdf" } } ], "alternates": { "markdown": "/en/publications/fandd/issues/2023/12/macroeconomics-of-artificial-intelligence-brynjolfsson-unger/index.md", "json": "/en/publications/fandd/issues/2023/12/macroeconomics-of-artificial-intelligence-brynjolfsson-unger/index.json", "bundleManifest": "/en/publications/fandd/issues/2023/12/macroeconomics-of-artificial-intelligence-brynjolfsson-unger/bundle-manifest.json" }, "generatedAtUtc": "2026-09-15T18:55:20.431Z" }