What is Artificial Intelligence and Machine Learning
Artificial Intelligence (AI) has many definitions but generally refers to computer systems that can mimic human intelligence through carrying out tasks such as learning and problem-solving. (1),(2) Machine learning (ML) is a subsection of this where computers learn for themselves from large datasets.[2] For example, in the context of climate change ML would be able to refine or discover links between data such as temperature and weather, whereas AI would build on these links to display automated warnings and related information. [2] Many approaches to AI require “training” in which it is shown labelled data as a base from which they “learn” to analyse new data. [4] AI is currently in use in many capacities, from analysing data of forests and coral reefs to determine the effects the climate is taking to collecting varying data to being used in Google searches. [1]
How would these computational methods impact global warming?
The IPCC has often emphasised the need to find solutions on a large scale, AI offers a way to look at data from many sectors like transport and electricity and suggest better alternatives or solutions to the emissions released from them. [4][7] AI’s ability to predict outcomes could be used for modelling and forecasting weather and managing supply and demand to allow for more efficient use. [7] This could also allow a role for AI in democratising electricity to make it more affordable as almost a billion people are currently without access. [1] Along with weather prediction, disaster prediction would be enhanced, allowing for a faster response time which could save thousands of lives. [7] AI could enable automatic monitoring of deforestation, analysing disasters and even accelerate scientific discoveries by suggesting new approaches to various projects. [11] The optimisation of public transportation and traffic would decrease their carbon footprints which contribute 29% of the carbon emissions in the US. All this not only benefits the climate but also could contribute trillions of dollars to the global economy in the next decade. [1]
Challenges and Problems
Several issues come with using AI, as earlier discussed AI often requires “training” thus as if the original data has a bias this could persist throughout the analysed data and may affect the decision that the AI comes to, possibly leading to discrimination. [4] In the case where AI attempts to understand individual behaviour to possibly find ways to persuade people in the direction that the AI determines beneficial for reducing climate change there are concerns as to whether this prevents free will. [4] Privacy issues will likely become apparent when using AI to limit emissions by looking at individuals or groups of people’s behaviours and their energy usage. [4] There is also the possibility that the increased carbon footprint from AI research may have a worse effect overall on the climate. [4]
[1] Stein, Amy L. “Artificial Intelligence and Climate Change.” Yale Journal on
Regulation, vol. 37, no. 3, Summer 2020, p. 890-939. HeinOnline. https://heinonline.org/HOL/Page?handle=hein.journals/yjor37&div=24&g_sent=1&casa_token=vwLnggj3LPcAAAAA:g9pQ1Pm3PeRUuMcaiFe4NGTLFlbPNDiO7yWoFslG-_8bJRwBeM18L2a5BR1jeRNnEoLoy_7_&collection=journals
[2] Chris Huntingford et al 2019 Environ. Res. Lett. 14 124007. [cited 16 August 2021];. Available from: https://iopscience.iop.org/article/10.1088/1748-9326/ab4e55/meta
[4] Cowls, Josh and Tsamados, Andreas and Taddeo, Mariarosaria and Floridi, Luciano, The AI Gambit — Leveraging Artificial Intelligence to Combat Climate Change: Opportunities, Challenges, and Recommendations (March 15, 2021). Available at SSRN: https://ssrn.com/abstract=3804983 or http://dx.doi.org/10.2139/ssrn.3804983 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3804983