10 June 2024
Can AI help fight climate change? We capture the discussion of a panel of experts as they seek to answer this question.
Last month, a panel of five academics and climate experts came together to discuss the opportunities and challenges of AI and sustainability, as part of the King’s Festival of Artificial Intelligence.
The Research Director of the Centre for Sustainable Business Professor Jonatan Pinkse kickstarted the event with a thought-provoking speech, where he explained how AI is already an integral part of managing greener energy solutions.
He was joined by Christer Hogstrand, Professor of Molecular Ecotoxicology, who shared his experience of chemical manufacturing and how AI could bring a wealth of benefits to the drug industry. Tamsin Edwards, Professor in Climate Change, discussed her research in quantifying the uncertainties of climate model predictions, and how AI has the power to reinvent weather prediction models.
Professor Frans Berkhout, Assistant Principal (King’s Climate and Sustainability), shared his team’s success in using digital copies of buildings to reduce energy emissions. Bringing an IT perspective to the table was Professor Luc Moreau, Head of the Department of Informatics. He talked about how AI is more like automation that captures data systematically, using advanced algorithms to create better predictions and real-time interventions.
The panel offered a variety of insights into the way AI is transforming their fields, but also the challenges that exist alongside the benefits. In this article, we share the event’s key takeaways and insights.
In Professor Jonatan Pinkse’s opening statement, he demonstrated how AI is already being used as an effective tool to combat the climate crisis. From wind turbines to solar panel farms, AI is integral to managing our growing network of greener energy solutions. It can speed up decision-making, accurately predict demand at peak times and balance the grid, all whilst ensuring cleaner energy is going to where it’s needed most.
On a smaller scale, this AI network is mirrored within the buildings of King’s College London. Heating is currently the university’s largest source of carbon emissions, as Professor Frans Berkhout pointed out, so his team have developed a ‘digital twin’ of the entire estate to manage energy consumption more effectively. “At King's, we operate about 150 buildings in central London, which consume huge amounts of energy,” Frans explained. “AI is absolutely central to managing our complex estate. We’ve used it to optimise our energy use and choose more ethical emissions.”
Looking ahead, AI will play an increasingly pivotal role as we adopt more renewable energy sources, including zero-carbon electricity-generating technologies.
From wind turbines to solar panel farms, AI is integral to managing our growing network of greener energy solutions."
Jonatan Pinkse, Research Director of the Centre for Sustainable Business
However, AI is clearly not restricted to greener energy solutions. It’s booming in every field imaginable – and as quickly as it’s becoming a solution, it also has the capacity to accelerate at an exponential rate and create even more issues around our privacy, data protection, and ethical use, to name a few.
As Jonatan summarised: “No matter how many good applications there might be for AI, there is the unfortunate reality that the bad outcomes can fast outpace these.”
The question of improving our healthcare was also discussed. Professor Christer Hogstrand shared some sobering facts from his research: “Currently, in the EU alone, there are 100,000 man-made chemicals in the laboratories, 80% still waiting to be tested for human use.” As these tests are resource-heavy and take up to 10 years, it makes drugs even more expensive and slow to market.
Christer’s team are now experimenting with AI to recognise the properties of chemicals and predict their viability, without the lengthy testing. This will give the team a chance to catch up with those 80,000 untested chemicals, and more importantly, AI could ensure that toxic chemicals are never produced in the first place.
However, Christer warned the audience that their AI model is trained based on past predictions, and that comes with its challenges: “What happens when the unpredictable happens, such as the terrible Thalidomide outbreak? We cannot train AI to predict the unknown.”
We cannot train AI to predict the unknown."
Christer Hogstrand, Professor of Molecular Ecotoxicology
This concern over the unknown element of AI was shared by Professor of Computer Science, Luc Moreau. As Luc highlighted in his opening statement, AI is like a black box, posing challenges in understanding both its decisions and its energy consumption.
Historically, the IT sector has been viewed as relatively ‘clean’ compared to transport or energy, but now there is a lack of knowledge about the true environmental impact of data centres. Luc explained how we must balance the benefits of technological advancements with their environmental implications, and how understanding this balance will be crucial for sustainable progress: “Technology alone isn’t the solution, it’s also a social problem.”
But there was a solution ready in Luc’s speech: education. “Societal and behavioural changes are necessary, and education plays a key role here,” he explained. “Universities are pivotal in researching and developing new techniques to address these issues, starting with incorporating sustainability more into our curricula.”
Technology alone isn’t the solution, it’s also a social problem.”
Professor Luc Moreau, Head of the Department of Informatics
Weather forecasts are fundamentally uncertain, so predicting a range of scenarios is crucial for decision-making, from warning the public about hazardous weather, to planning renewable energy use.
In Professor Tamsin Edwards’ speech, she used GenCast as an example of a new AI weather forecasting model that offers probabilistic data predictions. GenCast has greater skill and speed than existing supercomputing models and, when she visited a conference for a demonstration, it took just eight minutes to generate a comprehensive forecast.
Tamsin reiterated to the audience how critical AI could be not only in weather prediction, but also how both people and planet can benefit: “It’s clearly going to save computing energy, but more importantly, it makes weather forecasting more accessible to countries that don’t have state-of-the-art models.”
[AI] is clearly going to save computing energy and make weather forecasting more accessible.”
Tamsin Edwards, Professor of Climate Change
After the main discussion closed, the floor opened to questions from the audience. Unsurprisingly, the first question was around regulation, and how we can ensure technological advances are taken whilst mitigating the risks.
Professor Jonatan Pinkse responded truthfully: there are no answers yet. But, as Professor Luc Moreau had alluded to, education might be one answer. Around the world, there are academics and climate experts dedicating their time and resources into researching how we can achieve responsible AI and harness its power for a more sustainable future.
Clearly, AI can achieve remarkable feats in almost every sector its applied to, from healthcare and education to greener energy and climate prediction. But if there is one key takeaway from the panel, it is to proceed with AI with caution, as there are always two sides to the same coin.
This article was brought to you by the Centre for Sustainable Business. Find out more about their work here.
Explore artificial intelligence with King's researchers during this five-day public festival.
