Many of the people who care about climate change have decided that AI is a villain in the story—like Isengard from The Lord of the Rings, an insatiable, resource-guzzling factory complex that destroys the surrounding natural world to fuel its own growth. With the boom of sprawling data centers, their enormous appetite for power, and resulting strain on the electrical grid, there are certainly environmental considerations stemming from the industry’s rapid surge. But AI has helped predict protein structures, enable faster cancer detection, and improve weather prediction, to name a few breakthroughs. None of those fields decided the technology was too risky to try. They decided the problems they were tackling were too important to ignore a transformational capability at their disposal. Combating climate change is one of the world’s thorniest challenges. Climate tech can’t afford to be the field that declines to draw on what may be one of our time’s most powerful tools. While critical debates rage about AI safety and responsible development of AI infrastructure, we still need to ask how we can use AI to tackle climate change. From where I sit, as someone who helps the world understand its greenhouse gas emissions, the answer is clear: start with data. The world runs on satellite data You used satellite data today, probably before breakfast. Your weather app told you whether to carry an umbrella. Then, your maps app helped you avoid the annoying construction slowing down traffic on your route. For billions of people, satellite data is woven into daily life. But climate applications haven’t yet made that leap. The satellites that track emissions, deforestation, and ice melt produce data that is still highly scientific and technical. Traditionally, it requires specialists to turn it into something a decision-maker can use. Bridging that gap is where I think AI can do the most good for the climate: the translation of distilling technical information into something that people can act on. What that looks like today Methane emissions are a good test case. At GHGSat, our satellites observe millions of facilities every year. No team of analysts, however large, could comb through all that data and still alert operators about their plumes at a useful timescale. To combat this challenge, AI is built into our data processing. We use AI to estimate, from a range of factors, how likely each observation is to contain an emission, processing the most promising ones first. Then, our team confirms the emissions. What reaches a customer is no longer a mountain of imagery, but an alert about a specific emission at one of their facilities. That’s translation in practice. AI speeds up our pace, even as we’ve grown our satellite constellation and the resulting volume of data. But assessing the likelihood of methane plumes is just the beginning. The bigger prize is bringing sources together. If you want to understand what has happened to oil and gas emissions in Europe or Central Asia over the last five years, finding the answer means significant research: combining company self-reported emissions, regulatory records, and what satellites measure, then reconciling it all alongside geopolitical shifts and macroeconomic conditions. That type of work takes months of effort from entire research teams. AI can make that far easier, distilling the technical satellite data, and including it as one contextual input into a bigger answer for a policymaker or a journalist. The same pattern is showing up across other climate applications. Take deforestation. Start-ups and NGOs already provide timely alerts about tree cover loss—but the notification that trees are gone is just one piece. Local jurisdictions need to know the cause: a new road, a fire, or so on. To provide this context, the team behind it uses an AI method to classify alerts into different drivers, pointing a response towards the right intervention. Technology that could make this routine is only starting to emerge. Geospatial foundation models, large AI systems pre-trained on vast archives of satellite imagery, can be adapted to many tasks rather than built from scratch for each one. We’re at the beginning of that, and I’m very excited about where it goes. Translation is the first step, not the last Better data doesn’t fix anything by itself. A clear answer to “where are emissions coming from?” is only valuable if someone with authority acts on it. But accessible data is the start. The operator deciding on infrastructure investments, the regulator deciding where to focus, and financial institutions deciding how to insure, require the most fulsome data we can provide. We made satellite data effortless for maps and weather. Climate deserves the same. So what can climate tech leaders reconciling with AI use do today? Weigh the trade-offs of AI instead of opting out. Point it at a translation problem within your organization, where it can really add value. And when the results arrive, work to ensure that decision-makers will take action. Soon, policymakers will gather at COP31. Methane will be on the agenda again, as it has been at summit after summit, and the pledges will sound familiar. What’s different now is how much of the picture is measurable, and how much more of it can be understood by people who aren’t scientists. AI won’t solve the risks of a warming planet on its own. But if it turns satellite science into answers anyone can use, it removes one of the oldest excuses for inaction in the book: ignorance. After that, someone still has to decide to take action. The sooner more people can see and understand the problem, the better chances are that they will act on it. The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.
GHGSat CEO: AI doesn’t have to be a climate villain
Full Article
Original Source
Read the full article at Fortune →KhanList aggregates and links to publicly available news content. We do not host full articles from third-party sources. Always verify important information with original sources.