The author models the economic implications of two contrasting boundary scenarios for artificial intelligence (AI) development: restricting AI to purely assistive tools that augment human productivity versus enabling AI agents that can autonomously perform tasks at least as well as humans and replicate themselves. The analysis suggests that embracing AI agents could result in the economy growing faster than limiting AI to tools would.

Quantifying AI’s Economic Potential

Growth Differentials Between Assistive and Autonomous Development Scenarios

Tobias Sytsma

ResearchPublished Oct 8, 2025

The author of this analysis models the economic implications of two contrasting boundary scenarios for artificial intelligence (AI) development: restricting AI to purely assistive tools that augment human productivity versus enabling AI agents that can autonomously perform tasks at least as well as humans and replicate themselves. The Agent Scenario represents the potential emergence of artificial general intelligence capabilities, whereas the Tool World scenario represents AI progress as a variety of narrow, specialized systems that enhance human productivity but do not achieve general intelligence. These stylized scenarios are not meant to be interpreted as predictions; rather, they are meant to establish a variety of potential economic outcomes for policymakers, economic advisors, and researchers working at the intersection of technology policy and economic strategy. Using a calibrated endogenous growth model and Monte Carlo simulations from 2025 to 2045, the analysis suggests that embracing AI agents could result in the economy growing 3.8 percentage points faster annually, on average, than limiting AI to tools would. This difference compounds to make the Agent World economy nearly four times larger by 2045. Although the Agent World scenario assumes successful resolution of AI safety and alignment challenges, the results highlight the substantial economic incentives driving toward autonomous AI development and illustrate the economic trade-offs that are inherent to different AI development strategies.

Key Findings

  • Building on established economic growth theory and using Monte Carlo simulation methods, the results indicate that Agent World exceeds Tool World by an average of 3.8 percentage points in annual gross domestic product (GDP) growth. The median difference is 2.6 percentage points.
  • By 2045, the model suggests that Agent World’s gross GDP is approximately 3.6 times larger than Tool World’s GDP, assuming successful AI deployment without major transition costs or policy constraints. Some simulations suggest that Agent World GDP is several orders of magnitude larger.
  • Regression tree analysis of model simulations identifies the conversion rate from compute to AI agents and the compute-specific investment rate as the most important predictive factors for growth differentials.
  • Agent World economies with 90 percent or more AI-agent share of R&D workforce achieve 25-percentage-point or higher growth advantages over Tool World by 2045, while partial integration creates smaller improvements, suggesting that complete transformation unlocks qualitatively different growth regimes.
  • On average, the economic opportunity cost of forgoing Agent World development through 2045 would require nonmodeled benefits (such as safety, stability, or distributional advantages) that are equivalent to 1.2–2.8 times the existing annual GDP, depending on risk preference.

Topics

Document Details

Citation

Chicago Manual of Style

Sytsma, Tobias, Quantifying AI’s Economic Potential: Growth Differentials Between Assistive and Autonomous Development Scenarios. Santa Monica, CA: RAND Corporation, 2025. https://www.rand.org/pubs/research_reports/RRA4220-1.html.
BibTeX RIS

Research conducted by

This work was independently initiated and conducted within the Technology and Security Policy Center of RAND Global and Emerging Risks using income from operations and gifts from philanthropic supporters. A complete list of donors and funders is available at www.rand.org/TASP.

This publication is part of the RAND research report series. Research reports present research findings and objective analysis that address the challenges facing the public and private sectors. All RAND research reports undergo rigorous peer review to ensure high standards for research quality and objectivity.

This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited; linking directly to this product page is encouraged. Permission is required from RAND to reproduce, or reuse in another form, any of its research documents for commercial purposes. For information on reprint and reuse permissions, please visit www.rand.org/pubs/permissions.

RAND is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.