The Net Climate Impact of AI: Balancing Current Costs with Future Climate Benefits
Name
The_Net_Climate_Impact_of_AI__Balancing_Current_Costs_with_Future_Climate_Benefits___v2 (1).pdf
Description
Thesis PDF
Size
846.27 KB
Format
Adobe PDF
Checksum (MD5)
671707fa0062c2953e0498348c2af445
Author(s)
Turliuk, Jennifer
Advisor(s)
Sterman, John
Date Issued
May 2025
Publisher
Massachusetts Institute of Technology
Abstract
What is the net impact of artificial intelligence on climate change? Existing studies focus on AI's footprint, but few analyze AI's trade-offs. This paper develops a framework to quantify both the Greenhouse Gas (GHG) emissions and the climate change costs and benefits of AI systems, addressing the time value of carbon and the installed base of existing AI infrastructure. We examine the energy demands of AI, which are growing rapidly and threatening companies' net-zero commitments, while also analyzing AI's potential to enable emissions reductions through applications such as optimized energy systems, demand response, grid management, and electrification acceleration. This research introduces the Net Climate Impact Score (NCIS) of AI, a novel equation to calculate the net climate impact of AI technologies that considers both immediate emissions and potential future benefits, and provides a methodology for assessing AI projects holistically. We demonstrate that while current AI applications are predominantly emissions-intensive, strategic deployment focused on energy system transformation could potentially deliver net climate benefits within specific time frames and applications. However, improvements in energy efficiency and emissions reductions resulting from AI are, absent climate policy, likely to generate both direct and indirect rebound effects that could undermine the emissions reductions and reduce the climate benefits of AI. The research concludes with policy and industry recommendations that propose technological pathways that could maximize AI's positive impact while minimizing its environmental footprint.
MIT Department
Sloan School of Management
Terms of Use
Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0)
Copyright retained by author(s)
Persistent DSpace Link
