Microsoft's Chief Questions Officer Trevor Noah on how AI can help protect from extreme weather

Hey, everybody. Trevor Noah here. And I'm coming to you from the vibrant city of New Delhi, where today I'm at the headquarters of SEEDS, a nonprofit organization that is dedicated to weather-related disaster relief and preparedness. Today, I'm going to be joining the co-founder, Dr. Anshu Sharma, who’s going to help me answer the pressing question. What can we do to help the most vulnerable communities to both prepare themselves and recover from the effects of climate change? You can't find ans

wers without asking the right questions. This is The Prompt, with me, Trevor Noah. Anshu, thank you so much for taking the time. Thank you for being here with us. You know, I remember the first time I came to India. One of the most exciting programs I was introduced to was the SEEDS program. Everyone in the space of AI was speaking about it and it felt like SEEDS was doing something that really answered the question, what is the good of AI? So, before we get into that, what are the major disaste

rs that we're looking at? For the last two years, we've been focusing a lot on heat. Heat? It seems to be the biggest thing coming up globally. coming up globally. How is AI helping you determine the risk profile of individual homes in a country as populated as India? Satellite imagery is globally available. It's available every two days, so you can keep refreshing it. You can actually zoom in and see every single house in the world. That's about 2 billion houses. So, that's the data you have. A

nd that data has a lot of hidden information in it. Every rooftop that you see is almost like a QR code. It’s literally as unique as a fingerprint. You break that fingerprint of the house down into maybe 100 odd parameters. How are you doing it for each place when you say every single place is different? You know you have different QR codes. So, now you need what we call a rule engine. Okay. You have to tell AI that here is the data and then it just churns it impossibly fast, right. All right, s

o Anshu, help me understand what we're looking at. So, we are looking at East Delhi. We have a satellite image over here. Now in the summers, we were working around the issue of heat. Right. So, when we run this model on heat, this is the kind of first layer of output that you get. We install what we call data loggers, little devices that are recording temperature and humidity, literally every minute. We want to know what relief the person in a house gets during the night. gets during the night.

The maximum casualties happen on the hottest nights, not on the hottest days because your body does not get relief. You can handle a few hours of heat, but you can't handle heat that just stays up. So then we start getting this kind of imagery that, hey, if the map is telling you that there's going to be a heat wave two days from now, now we know which houses will be struggling. The dream is that it eventually becomes something as simple as maps that you put on your phone and be able to see thi

s. But right now, this is a dashboard that we take to governments. So, you could choose your hazard. You are in Delhi, you could look at heatwaves, and you get individual scores for houses and clusters. So, you're warning people right now. You say to them, listen, your home is in danger of reaching a critical heat. Exactly. You would want to evacuate. You wouldn't want to be in that home. If there is a heat wave warning, we can actually tell you how hot it will be inside your house. We can tell

you what actions to take immediately. And 80% of the heat that a house gains comes from the roof. So, if you can insulate the roof, creating a kind of a buffer, we are able to reduce the indoor temperature by 12 degrees Celsius. That is amazing. It's a life-saving change, right? So AI’s a very clever tool, but it works in the background. The partnership is between technology and people on the ground. That's the stuff where we believe magic will lie. We are only seeing glimpses of it right now. A

nshu, thank you for the time today. Congratulations on everything you're doing and thank you once again for spending time with me. Thank you, Trevor. Thank you for your attention and amplifying our work.