On March 4, 2025, experts from various fields from Columbia University gathered at the rapidly growing field of artificial intelligence at Columbia University. The summit covers topics ranging from healthcare, business and policy to science, engineering and humanities, thus providing a 360-degree perspective on the transformative impact of AI on society.
From Chaos to Code: How AI tames the afternoon lessons of climate crisis, conducted by climate school researchers, describe how AI becomes a powerful tool for climate science, disaster preparation, and across interconnected systems. Please continue reading the highlights in the conversation or watch the video below.
Speaker
Beginner remarks: David SandalowSchool of International and Public Affairs
»View all Colombia AI Summit Videos
The power of AI
“There are many different changes [AI] Possible benefits. What I am particularly excited about is in material innovation… When Thomas Edison invented modern light bulbs 150 years ago, he spent months testing different types of elements or materials, running electricity through them to see how much light and heat will be generated. He finally came up with the solution he thought was the best. Today, we can use modern AI tools to simulate a million interactions in one second. This ability gives us two benefits. One of them is that we can test materials that do not actually exist. Then, if it looks like they will be beneficial in some application, we can make them and see if they work. Then we can choose faster. I think the potential of AI to help us accelerate the pace of energy innovation is beyond, and we need to figure out how to mobilize various tools to do that. ” –David Sandalow
“What’s happening now is AI [weather prediction] Models – models that didn’t exist in nature five years ago – become like [traditional] Physics-based model. AI models have a more or less explicit understanding of atmospheric physics. They don’t know about energy or motivation or any of them. Like all other AIs, they are trained in data. You’ll give them a lot of historical weather data and know what happened in the past and then see what happened later, and they find patterns in mysterious ways of machine learning and AI. So, like other AIs, training is very expensive, but as David says, predictions are really cheap to run in real time, and they become very powerful in a short time from everywhere and in a very short time. ” –Adam Sobel
“I think equality is crucial. Because in the past, when there was something wrong with the power system (which is exactly what happened during the winter storm Uri), power operators were unable to provide enough energy and they needed to reduce demand. I think Texas cut off about 30% of demand during this period (because they lost a lot of generations) to keep the rest of the system executed. But, If you consider an energy system (basically a branch network structure), the easiest way to reduce demand is to cut off branches. However, usually or historically those vulnerable communities living in branch areas are often less recent infrastructure… On the other hand, with new deployments of batteries and solar PVs, we are seeing affluent communities deployed, such as Tesla Powerwalls, such as Tesla Powerwall, can provide homes. We should start rethinking how to reduce demand in emergencies and considering that certain homes or buildings can supply themselves through the microgrid, but we also have vulnerable communities. In these cases, it can be helpful to make recommendations for power system operators, make quick decisions, and sometimes even simulate different situations. In these areas I think AI has shown great hope.” –Bolun Xu
“As a behavioral ecologist, I’m very interested in how individual organisms respond to environmental changes. And I think to start making real predictions, we need to understand changes at the individual level in order to be able to predict which populations and ecosystems will respond to climate change or land use changes. I think that’s [species identification] The technology can be very useful. We’re moving from individual species recognition to individual identification using images from camera traps or from recording pods … We can use images to capture individual zebras—I think that was the first species because they have almost a barcode, like a fingerprint—and you can make really good population projects and follow those animals and see where they go, so to see what happens during droughts or periods of land use change … I think as we move to projects, we can scale up from actual individual level variation and differences to the population, to the ecosystem, and make better predictions about how organisms cope with changes. ” –Dustin Rubinstein
“This is a truly unique opportunity to build methods with AI to determine the best adaptation and resilience strategies for infrastructure targeting weather and climate-related hazards. This problem is extremely challenging and highly multidisciplinary and it will involve contributions from engineering, physical science, social sciences and many other fields. The only way to get close to it is through a novel Monte Carlo approach, including stochastic optimization. AI will help us integrate all of these different disciplines to ultimately solve the problem.” –George Deodatis
The challenge of AI
“We all know that predicting hurricanes, especially in terms of climate, is this hurricane that will happen in the next 5, 10 or 20 years, and what is the frequency and the expected intensity, is very difficult. So far, we have some models, probabilistic models, that describe this uncertainty. However, these models assume that there is a fixed climate, which This climate is unchanging. We now know that the climate is changing significantly, so these probabilistic models of extreme events will change over time. This makes the problem more challenging. Therefore, we rely on artificial intelligence methods to determine that certain models can quantify the occurrence and evolution of these extreme events in a probabilistic way.” –George Deodatis
“If you look at the food system, you see a lot of inequality. Little shepherds, shepherds, the custodians of the ecosystem and the small farmers of our food system, the indigenous population and people, usually lag behind better data and decisions in this game. I think we have to use these technologies very carefully, especially because we have so many small farmers who are often left behind. So democratizing data around AI will be so important.” –Jessica Fanzo
“With AI moving so fast, it’s possible to use something people don’t understand, I think there’s a tension between the public and the private right now… There’s now the White House move to privatize the weather service, which has been discussed for years, but now there’s a real threat…new…new…new… [AI models] From the private sector, but it still depends entirely on the public sector infrastructure of the underlying data and the physical model that works in many backgrounds. So there is a lot of tension here, and I think we rely on the real danger of destroying things in infrastructure that we rely on to ensure people are safe. ” –Adam Sobel
“I think about 30% of California’s daily power generation capacity is now from giant batteries. It’s fair to say that many of these batteries are now operated by AI, of course, under human monitoring … in many cases we can’t explain it in many cases. [what the AI is doing]making power system operators worried. It’s really about understanding transparency and how to adjust that AI, which I think is a real focus for many power system operators. ” –Bolun Xu
“We know these models depend on the data we train them, and I think most of the people here are talking about a species: humans. What happens when we train a model of a species or a family of species, and then we cut it off Go around the corner and try to project it with other species? I think that’s a real problem. When the training data is different from different things, can we improve our model so that we can be able to work with another ecosystem or other species? ” –Dustin Rubinstein
The Colombia AI Summit is organized by Colombia AI, a program aimed at promoting Colombia’s work in artificial intelligence through courses, courses, events, digital tools, and more. Columbia AI is a university-wide effort through a partnership between the Institute of Data Science, Columbia Engineering and Executive Vice President..
*Highlights have been edited as clear
