Opinion

AI can help heal
us heal the planet

Anil Madhavapeddy

Professor Anil Madhavapeddy

Professor Anil Madhavapeddy

AI can do in seconds what might take a team of experts a year. This is why we must harness it to reverse the damage we’ve done to the planet. Anil Madhavapeddy explains.

We need to act fast to mitigate the impacts of climate change, and to protect and restore biodiversity. There’s incredible potential in using AI to augment our work. It enables us to do things much more quickly – it’s like giving humans global knowledge superpowers!

It can turbocharge human capabilities, simulate complicated systems and search through vast amounts of data. It would help us make rapid progress in reversing the damage we’ve done to the planet with well-targeted interventions, while continuing to supply human needs.

Of course, this comes with risks of runaway systems causing harm, so everything we do must include a ‘human in the loop’ to guide what’s going on. AI systems have no capability for nuanced judgement.

Humans are generating vast amounts of information about our natural world. The imaging data alone spans every scale, from satellite and drone images of countries and landscapes, to land-based photographs of species and habitats, to microscopic observations of life. Alongside the visuals, conservation and climate scientists and practitioners are publishing an ever-increasing amount of written information on their ideas, experiments and real-world trials.

Imagine having access to all of this, plus razor-sharp climate models, available at your fingertips - with answers about any real or imagined situation available in seconds.

The Holy Grail is to combine all this observational data with all knowledge-based data from the whole of humanity and generate evidence-driven insights to accelerate our collective healing of the planet.

AI algorithms, searching and analysing the data, could help empower decision-makers to be confident that they’re making the best choices.

Ultimately, we should be able to create AI ‘Co-Pilots’ for policy-makers, to help them make decisions about all sorts of things in the best interests of our planet – whether a new development in Cambridge is a good idea, for example. AI could quickly create a referenced, in-depth report on anything a policy-maker wants to know, and forecast what will happen as a result of any specific decision.

Achieving its potential

There are currently three barriers to achieving this promising vision: access to enough hardware, energy and data.

Data is fuel for AI – but it’s been an enormous challenge getting hold of enough of it – particularly accessing published journal papers. The government has a desire to create a National Data Library, which is a great idea, because it would allow us to access huge amounts of existing knowledge and run it through AI algorithms while preserving privacy needs. Right now, the data is scattered and difficult to securely access for researchers and policymakers.

We also don’t have enough hardware. We need more GPUs – graphics processing units – they cost around £40,000 each and we need hundreds of thousands of hours of GPU time to unlock the scale required for modern learning algorithms.

And on energy, the fact that AI uses huge amounts of it is a big concern, but there have been recent research advances in the core AI approaches to make our energy expenditure much more efficient. Simulating the planet is also what's called a 'root node problem' in that it is the beginning of a continuously 'branching tree' of other computational possibilities that will unlock ways to improve human lives.

If the barriers can be overcome, then the potential for AI to help us address the climate and biodiversity crises is huge. Through collaborative efforts across departments, Cambridge is harnessing the power of AI to work alongside some of the world’s brightest minds. There has never been a greater opportunity to develop solutions for our planet’s future – and to help rebalance the relationship between humans and nature across the world.

Anil Madhavapeddy is Professor of Planetary Computing in the Department of Computer Science and Technology.

Published 5 April 2025

Photography: Lloyd Mann
Interview: Jacqueline Garget

The text in this work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License