Changing Tides with AI-Powered Climate Change Analysis

Climate change represents one of the most important challenges facing humanity, and its implications extend to 2099 and beyond. However, detailed and applicable weather forecasts over many centuries are rare. This paper seeks to harness sstate-of-the-art artificial intelligence (AI) and machine learning (ML) capabilities for hundreds of years of time series climate analysis. Our method extrapolates climate numerical simulators trained to estimate future temperature scenarios based on socioeconomic emission patterns. We provide performance values in terms of out-of-sample forecasting functions and quantify emissions uncertainty as predictions extend further into the 22nd/23rd century and beyond. Additionally, weexplorethepotentialandpitfalls of AI when incorporating long-tail risks such as climate change. Our contribution provides a unique long-term perspectiveidentification of climate milestones, key decision points, and analysis of climate sensitivity over a multi-century period. We highlight ethical considerations of uncertainty diffusion and discuss ways to improve understanding of climate change likely to be experienced by very distant descendants. This paper aims to broaden and enrich conversation about humanity's responsibility across generations beyond our immediate descendants facing profound future climate concerns