Climate change and rapid technological transformation are some of the twenty-first century’s most pressing issues, with artificial intelligence (AI) playing a potentially vital role connecting the two. So, what specific climate change challenges can AI address? Sims pauses and takes a deep breath. “Well, there are many, across various domains. AI can help us to solve challenges in energy, forestry and land use, industry, agriculture … the list is long.” Ultimately, to address climate change, we have to reduce emissions to zero across the whole of the economy. Pete adds, “The power sector, the transport sector, heavy industry and buildings … these are the core sectors that are really critical. And there’s a whole ton of examples where AI is already being useful to help those sectors and systems transition.”
AI’s power to accelerate climate action lies in its ability to ingest huge amounts of data, to optimise complex systems, and to improve forecasting. “AI can be really helpful in terms of distilling data into useful insights,” Pete outlines. “That can be collecting and exploring satellite data identifying areas at risk of illegal logging, or data looking at installations that are emitting methane when they shouldn’t be.”
The duo point to a number of current-use cases that they shared in a recent report with the Global Partnership on Artificial Intelligence, highlighting that this isn’t just a hope for the future (helloworld.cc/aireport). The UK National Grid, for example, already uses AI to optimise electricity demand forecasts and renewable supply. Aionics, a US-based start-up, uses machine learning to cut down on the research and development stage of building batteries, which can be used for electric cars. And Kuzi, a tool developed by a Kenyan company, aggregates various data and uses AI to make predictions about locust breeding locations and migration routes, to mitigate potentially devastating locust outbreaks.
The potential for AI to support action on climate change is clearly high, but it’s a powerful tool, so it’s vital to ensure it’s being used in the right way. “This is a really complex space and even with the best of intentions, applications can sometimes drive suboptimal outcomes,” Sims warns. “We must remember that AI is domain-agnostic — it’s just a technology, so it can be used in oil and gas exploration and extraction, or other applications that are not good for the environment. There are also the knock-on effects of AI — autonomous vehicles, for instance, offer efficiency gains, but maybe also bring more drivers onto the roads, so more emissions. We must understand the many implications of what we build and make responsible choices.”
Responsibility is the key word here, and an important role of the Centre is bringing together people with different expertise and facilitating discussions. Sims explains, “Having that collaboration and ongoing conversation allows us to drive towards the more climate-positive outcomes [of AI], and to make folks aware of the applications that do run counter to climate goals.” Pete adds, “Responsible AI is on everyone — it’s something that everyone needs to be thinking about, whether that’s the data scientists developing tools, or the governments who are setting the frameworks and regulatory environments in which these tools operate, or the businesses and companies that use this technology in their operations.”
As time goes on, the need to address climate change is becoming more urgent, and as Pete puts it, “There’s going to be increasing tailwinds for policymakers, companies and everyone, really, to use whatever tools we have in our toolbox to accelerate the transition [to net zero] as fast as we can.” AI-readiness is the next big task on the Centre’s agenda, and they’ll be working with the Global Partnership on Artificial Intelligence again to educate the sectors vital in the fight against climate change (energy, transport, agriculture, and so on) to highlight what’s possible with AI and what’s needed to become AI-ready. “These cross-domain conversations are so important,” Sims stresses. “There’s a lot of curiosity within those industries, so we need to be able to communicate the potential effectively.”
Ensuring that the right skills are being developed in schools, then, is going to be of the utmost importance. “The more we are able to upskill in cross-disciplinary education, the better off we’ll be in the future. One of the gaps we have right now is people who have deep domain knowledge as well as machine learning or computer science backgrounds,” Sims explains. “If you’re able to understand the domain problem and also help the domain experts translate that into computational language that algorithms understand … that is going to be such a vital skill. If students are interested in this space, though, get involved now — they shouldn’t feel it’s something they should wait until they’ve got a PhD to get stuck into!” It may not be a silver bullet, but it’s clear that there will be plenty more coffees and conversations to come about this powerful tool.
For more information about the Centre for AI & Climate, visit c-ai-c.org.