You could argue that AI entered the public consciousness in 1997 when the reigning world chess champion Gary Kasparov lost a game of chess to IBM’s Deep Blue, a chess playing computer program. Nearly thirty years previously, the renowned British mathematician Alan Turing (who cracked Germany’s Enigma code during World War II) wrote a paper called Computing Machinery and Intelligence, which examined how to build ‘intelligent’ machines. He reasoned that if humans could use all available information to make decisions, then surely machines could do the same thing. Of course, what hampered Turing, and others doing similar work, was that computers weren’t powerful (or inexpensive) enough at that time. Luckily, Moore’s Law (the idea that computers’ memory and speed doubles every year) meant that computer capabilities would eventually catch up with, and even surpass, human imagination.

Of course, today AI is everywhere: from spam filters to drones, it has become an accepted, almost invisible part of modern life. So, given that we are living in an era of Big Data and Machine Learning, how can AI help us in the fight to take climate action? Can we use computers to help us reduce emissions and help make, for example, our energy systems more efficient?

Stephanie Sy is the founder and CEO of Thinking Machines, a data science start-up with the goal of making artificial intelligence systems work for humans. A Stanford graduate, she moved back to her home country of the Philippines to grow her company which “solves high impact problems for large organizations across South East Asia.”

Big data, big opportunities

She’s optimistic about the role AI can play with regards to climate action. “If we look at South East Asia, a lot of emissions are a result of development – land development and deforestation, and the use of fossil fuels for energy,” she says. “So how can AI minimize deforestation and flip that into reforestation, and allow for efficient carbon-neutral land development and improved energy efficiency? If you look at the UN’s Sustainable Development Goals, both the words ‘sustainable’ and ‘development’ are in there. We want to grow sustainably but we also want to develop so people don’t stay in poverty. How do we bring people to a good quality of life while doing so sustainably? And that’s where AI can help.”

The good news is that governments are now starting to utilize AI across the region. “AI is having an impact in South East Asia in monitoring deforestation and things like illegal mining,” she says.

AI – essentially a computer algorithm – can take millions of pieces of data (everything from previous weather reports to satellite imagery) and process this data and create patterns that mimic the human brain in order to make decisions, and predict future behaviors.

“Traditionally, those data points are really hard to get, and it doesn’t really help if you find out six months after the fact that a thousand acres of forest has been burned down. What you want is a safe, real-time monitoring system, so when you start to see illegal logging happening, you can immediately intervene without having to wait for annual cycles.”

Sy is also seeing increased usage of AI in the private sector, particularly regards to climate change mitigation. “We see insurance companies and agricultural companies trying to use AI to predict flood risks and crop failures as it’s a place where better climate outcomes and better business outcomes go together,” she says.

This is echoed by UNEP, which helped organize the AI For the Planet virtual conference in February. UNEP’s Executive Director, Inger Andersen says that AI can help ensure we get to better climate outcomes quicker. “How do we use digital solutions to drive sustainability and to create a world that is circulator, regenerative and inclusive and where we know how we are tracking and measuring where we are falling behind? [So, we] are just beginning to support and scale environmental change through the digital architecture,” she says. Indeed, the general consensus of the Summit was that the use of AI and big data in areas such as buildings, energy, agriculture and manufacturing, can result in a “10 to 20 per cent reduction in global carbon emissions by 2030.”

AI is not without its issues however; all our digital activity has a carbon footprint, and certain AI models (again, think of a huge computer software program) use huge amounts of electricity. A University of Massachusetts research paper published in 2019 reported that training (inputting data so the AI can learn patterns) a large AI model can result in more than 600,000 lbs of C02 emissions, five times the amount produced by the average car over its lifetime.

“The models [referred to in the University of Massachusetts paper] are one-off massive models by big American corporations,” Sy says. “So when someone talks about five cars worth of emissions, they are talking about some of the most complex, largest ‘pre-trained’ models in the world,” she adds.

However what Sy and other data scientists are doing is ‘piggybacking’ on the huge models already created, and fine-tuning them.

“So, we don’t generate all of those emissions all over again, we add just a little bit of emissions. You only need to train those massive models once, and then everyone else can build on top of that. I would be much more concerned about cryptocurrency and blockchain where its linear and all the emissions are dumped one-time up front,” Sy adds.

AI efficiencies

And there are many examples of projects where AI has had an impact on reducing carbon emissions. A group of Cornell University researchers and computer scientists developed an AI model to find hydropower dam sites in the Amazon Basin that would produce the lowest amount of carbon emissions. AI can also help farmers monitor crop yields, which reduces the need for water and fertilizer and reduces waste through demand forecasting.

Last year, a UN Climate Change event focused on how technological developments in satellites, artificial intelligence and big data could work towards an international data set to support implementation of the Paris Agreement

In 2018, the United Nations’ AI For Good Global Summit brought together 30 UN agencies, as well as AI innovators and public and private decision makers to generate AI strategies and support projects to accelerate progress towards the UN Sustainable Development Goals (SDGs). Sy for one is optimistic about AI’s role in climate action.

“I think moving forward we will see AI strongly influence climate action in terms of efficient energy generation and storage, especially in providing pathways in renewables and microgrids,” Sy says.

“I would like to see more power in [Asia] coming from renewables, but the problem with wind, solar and tidal power is they are all highly variable and you can’t run a factory if you are going to run out of power intermittently throughout the day. So AI can help predict power fluctuations and help route power better to connect renewable power sources to national grids, and help to do load balancing. It hasn’t been done yet but I am really bullish about AI’s ability to have an effect in this energy space.”

The future of AI is complex, and while there are lots of opportunities around how it interacts with climate action, there are lots of concerns too. Key is placing climate action at the centre of the decision-making process and ensuring that the intelligent machines Alan Turing dreamt off are used to help, rather than hinder humanity.