Welcome to AI and Climate Change, the second course in the AI for Good specialization. In this course, you'll see how AI can be part of the solution for challenges that we face with climate change. Climate change is already impacting human populations and natural ecosystems through droughts, floods, wildfires, rising sea levels, and other adverse phenomena. And in this course, you'll work through case studies and see how teams around the world are approaching the problems presented by climate change. And you'll see where AI can be applied to applications in renewable energy and biodiversity monitoring. I'm thrilled that, once again, we have as your instructor, Robert Monach. Thanks, Andrew. It's great to be here. And I think this is one area where you might have more expertise than me. I think you've done some work in AI for climate change in the past. Yeah, in fact, my collaborators and I wrote quite a few papers on using satellite imagery to understand drivers of deforestation and methane emissions and so on. So I think climate change, in addition to being important to every person, is also the technical aspects that are near and dear to my heart. Oh, that's great. Yeah, that's very similar to the use cases that we'll have here today. So we'll talk about biodiversity monitoring and wind power forecasting. Both of these are very important topics in understanding how we can reduce our reliance on fossil fuels and monitor the impact of climate change on the biosphere. In fact, wind power forecasting is one of those techniques where AI is already having a significant impact on the operation of wind farms and on planning on electrical grids. It's really having an impact. And as an AI person, I feel proud of our field for having that impact. Yeah, it's incredibly exciting. I think we've got to a point now where producing electricity and methods with a low impact on fossil fuels like wind and solar is becoming very effective. However, the prediction component is still very hard. We don't always know how sunny it's going to be or what the wind is going to be like. And these are the kind of problems that lend themselves really well to machine learning solutions. Because you're essentially taking a large number of signals, more than you could write a rule-based human system to really combine. And you can use machine learning to get better predictions. And even a few percentage improvement in wind or solar can have incredibly large impacts in things like the reduction amount of carbon emissions. Yeah, because they can better forecast at a certain time in the next few hours, the next few days, how much solar, how much wind. They're going to decide to ramp up or down higher polluting plants and just optimize the power grid. Maybe we can charge people's electric vehicles at a better time of day. So this information is flowing through the grid in a way that I find extremely exciting. It's just making the power grid more efficient. Yeah, I think that's a really good way to look at it. The power grids built on fossil fuels will often only have a variance of a couple of percent, any higher or lower. And people start getting blackouts or brownouts. And so we're really rethinking what power grids look like when we can have distributed power from people's batteries in their homes or in their cars. And power that can be shared across multiple grids based on what renewable energies are able to produce power at what times. And so in this course, you learn more about that aspect of trying to address climate change. And then the second major project is biodiversity monitoring. It's also important to think through how climate change is already affecting biodiversity. Do you say more about that project? Yeah, the second project is really fun. So we are looking at images from a national park in South Africa, trying to gauge the number of animals. So this can tell us what is the environmental impact or the other human impacts which are reducing the number of animals in a given location. And I think it's a good example, both in the positive and negative. This is an area where we can very effectively deploy existing machine learning technologies. But also we have to ask ourselves, well, what is the most important thing to look at? So for a lot of biodiversity monitoring, we care about plants or bacteria. But that's harder to do. So I think this is fun. We're looking at what I've heard being called charismatic megafauna. So it's fun looking at these animals and we have good object detection methods that can be used for that. But at the same time, we have to think, well, OK, there are reasons that we can detect large animals better than we can detect small ones with current machine learning technologies. So it's also a good way to think about where the gaps are today. And hopefully that will inspire people to think about ways of addressing those gaps in the future. So I think in the previous course, you started to see a systematic framework for how to address complex problems. And so in this course, you see that framework continue to be refined and adapted to these two really exciting applications of wind power forecasting as well as biodiversity monitoring. And so with that, I'm excited to have you learn about these topics. Please jump in and let's go on to the next video.