Code and Data: Soaring Carbon Emissions of AI Threaten Climate Goals and Its Social Costs

The files hosted on this Zenodo page are used to conduct the analysis needed for the research article “Soaring Carbon Emissions of AI Threaten Climate Goals and Its Social Costs”. 1. green500_top_data.xlsx This dataset contains organised data for globally leading supercomputers (https://top500.org/lists/green500/), including information on energy efficiencies, locations, industries etc. 2. country_spec.xlsx This spreadsheet contains IPCC Annex labelling for countries included in this research. 3. Epoch_AI_data.xlsx This dataset contains organised data for globally leading AI models released (https://epoch.ai/data/notable-ai-models), including information on number of parameters, computational consumption, industries etc. 4. IEA Energy Prices - Yearly.TXT This dataset contains organised data for energy prices of different countries by IEA. 5. electricity mix.xlsx This dataset contains organised data for electricity energy mix of different countries by IEA. 6. electricity ef.csv This dataset contains organised data for the emission factors of different electricity sources for countries by IEA. 7. 30_50_scenario_data.xlsx This spreadsheet contains the setting of parameters for scenario setting applied in this research. 8. AIModelEmission_v12.m This Matlab script is the main code used to generate the result of this research. 9. SCCvsAICO2_v2.m This Matlab script is the code used to calculate the future social carbon cost of AI's emissions in this research. 10. USEPA_sc_co2.xlsx This spreadsheet contains the dataset of social carbon cost simulated by the U.S. Environmental Protection Agency 11. forecast literature v4.xlsx This spreadsheet contains the growth rate projection of energy consumption by AI in multiple literatures