Videos of sudden rain and storms causing floods all over Dubai are circulating like wildfire. Sometimes, nature likes to show us who is the boss. While humans tamed the vast desert to build a beautiful and modern city, they could not tame nature. If at all, the carbon footprint of our towns is wreaking havoc on climate patterns, thereby exacerbating nature’s fury. The videos are difficult to watch, so my thoughts and prayers are with those who were impacted. Looks like there has been only one loss of life, but severe disaster to property and infrastructure.

Watching all those videos made me think about a system for providing early warning to city residents in case of such acts of god. In scenarios of disasters like these, even seconds count. In one of my posts a few months ago, “Birds, Earthquakes and Intermediary Fallacy“, I wrote about how AI can help decipher some early markers in the case of earthquakes. Fortunately, climate wraths like storms are a bit easier to detect in the near term than earthquakes.

In this specific case, a particular pattern of climate formed hours before it all began. Climate forecasting has been a thing for centuries. Obviously, science has made it much more precise. However, long-term and mid-term forecasting are still prone to errors. The best accuracy, like every other form of forecasting, is in the near term. But sometimes, even near-terms are a complete miss, as nature can sometimes change its mood. But even in those scenarios, even a 30-second warning may save many precious lives.

Upon researching the specifics of the climate system that formed before the deluge, it was evident that an AI algorithm, trained using the massive climate-related dataset that humanity has amassed over centuries, could have predicted the severity of what was coming. Even a 30-minute warning could have allowed many to take some precautionary measures, like getting off the streets, returning home, and closing commercial establishments. In evacuation scenarios, a decent amount of evacuation could be conducted. The crux is that even if the warning is last minute, it can still help minimize damage to some extent.

In 2021, before I moved to the Chicago suburbs, a tornado hit the street where I currently live without warning. It flattened a few houses and damaged several, including the one I live in. Many people were injured, some spending months in ICU and rehab. Human injuries could have been avoided if there had been even 15 seconds of early warning. That would have been enough for many to shelter in their basements and crawlspaces. Some residents were badly injured while they were sleeping.

Tornados are trickier. However, upon research, it was evident that you can build an AI-enabled system to warn residents at least 30 seconds in advance, even for the most unexpected. An algorithm could chart possible paths and then leverage existing infrastructure that towns have, like warning systems, to warn residents in areas that may get impacted. A tornado may take only one of those possible paths as it moves, but it is better to be safe than sorry.

While doing this research, I recalled the cloudburst disasters that happen in India every few years. I researched that as well, and it looks like there are some last-minute markers that an AI algorithm can be trained on to provide a warning and save as many lives as possible.

Every capability needed to design and train AI-enabled solutions like these is now available. Smart people among us have recently made a lot of progress in Artificial Intelligence, and we now have capabilities that can be tapped to address many challenges in both the public and private sectors. This is one of them. Let us start leveraging the power that has been made available to use for innovative solutions to persistent problems.