You can enable subtitles (captions) in the video player
Humans are responsible for climate change. Now, they're investing their hopes and money in artificial intelligence to help tackle it. AI can play a crucial role in analysing the increasingly vast and complicated weather data gathered from around the world, monitor and mitigate emissions, and optimise transportation networks and agriculture processes.
Microsoft has put $50mn into its AI for Earth initiative, which grants companies access to AI to support projects that analyse climate or energy data. The benefits for Microsoft include road testing its tech, and moving closer to start-ups it might want to buy or invest in. Google is deploying its DeepMind AI to green its own significant energy consumption in its data centres. And IBM is using machine learning to optimise flight routes and reduce fuel consumption in the aviation industry.
An increasing number of start-ups are also leveraging AI to help businesses and municipalities plan and prepare for the impact of extreme weather, or to make transportation, such as cargo shipping, which is a significant cause of global pollution, less carbon-intensive, by revising routes and fuel consumption.
AI's strength is its ability to highlight inefficiencies, paving the way for incremental, but vital optimisation across various industries. In agriculture, AI could optimise fertiliser use by using data for factors such as weather conditions and nitrogen levels in the soil.
Since weather fundamentally drives both variable generation and electricity demand, advanced climate modelling using AI could improve efficiency across the electricity grid. For example, the US government has invested $4.5bn in a smart energy grid that uses AI to measure how much is needed, and when.
The UN predicts that AI will also be fundamental in driving sustainable consumer choices calculating the environmental footprint of a product, and enabling businesses, such as Amazon or Alibaba, to nudge their online customers toward greener choices.
The use of AI comes with challenges, too. There are ethical concerns surrounding machine learning, and the amount of data and public surveillance that it might require. And its analysis is not always reliable. But the biggest challenge facing AI in its role as climate warrior, is its own growing carbon footprint.
Storage and processing needed to fully train a large algorithm can consume as much as 284,000kg of carbon dioxide. That's nearly five times the life-cycle emissions of an American car. Researchers say that transparency, such as open reporting on the carbon footprint of AI models, and the weeding out of inefficient systems, is the key to ensuring that AI is a help, rather than a hindrance in the battle against climate change.
