An artificial intelligence found that we will reach the critical threshold of global warming much sooner than expected. In a study published in Proceedings of the National Academy of Sciences, researchers used a machine learning model to find out when we will cross the 1.5 degree Celsius barrier and the results are alarming. According to scientists, the Earth is about to catch up, even if measures are applied to reduce climate change in the short term.
Noah S. Diffenbaugh, a climatologist at Stanford University, and Elizabeth A. Barnes, a professor in the Department of Atmospheric Science at Colorado State University, relied on artificial neural networks (ANNs). Using a spatial pattern of historical temperature observations, artificial intelligence estimates that we will reach 1.5 degrees between 2033 and 2035.
The predictions were made considering three scenarios of greenhouse gas emissions. Under these parameters, the researchers also found that the Earth will reach 2 degrees of warming between 2050 and 2054. The artificial intelligence estimates are consistent with the assessment of the Intergovernmental Panel on Climate Change (IPCC). The results of the Sixth Assessment Report (AR6) predict that we will cross the 1.5 degree Celsius threshold in the early 2030s.
In the case of 2 degrees, there is a difference when compared with some forecasts that placed it at the end of this century. The researchers mention that the AI estimates coincide with the AR6 assessment, which considers 2041 and 2060 as the year in which we would reach this temperature in the high and medium emissions scenarios, respectively. The study also emphasizes that not all climate models reach 2 degrees Celsius in a low scenario.
How reliable is artificial intelligence in predicting climate change?
Diffenbaugh and Barnes say that the use of artificial intelligence provides a unique data-driven approach. The neural network-based model accurately predicts the timing of global warming from temperature maps. It is worth mentioning that these networks learn from climate models, so these are required to have some truth in their representation of the real world.
One of the drawbacks of the AI learning process is that there can be biases and errors in the data. To avoid this, the neural networks were based on multiple climate models and considered different forcing scenarios. The researchers say that when comparing the results with previous evaluations there may be variations.
Given the existing substantial evidence of accelerating risks to natural and human systems at 1.5°C and 2°C, our results provide further evidence of high-impact climate change over the next three decades.
Although the results are alarming, Diffenbaugh is hopeful in the changes brought about by the Paris Agreement. Global warming would have catastrophic effects worldwide, beginning with a rise in sea level. Experts estimate that with 1.5 degrees we would experience growth of 48 centimeters by the end of this century, while at 2 degrees, levels would reach 56 centimeters.
Added to flooding and erosion in coastal cities, heat waves would become commonplace. At 1.5 degrees, Europe would have a 47% chance of experiencing a severe wave every year. However, if the AI model is correct and we reach 2 degrees in 2050, it will be almost a given that we would suffer every summer with high temperatures that would take the lives of many people.
Both the AI and IPCC predictions state that we will get to 2 degrees this century. The question is when.