Forbes recently released an article titled “Is Artificial Intelligence Dangerous? 6 AI Risks Everyone Should Know About.”

Brilliant, another mainstream news outlet demonising the applications of AI. It appears that these stories are ten a penny in 2018, the notion that AI is dangerous and should be feared is fast becoming monotonous dogma; a concept recycled again and again. I’m not saying that AI does not need to be approached with caution, I am simply highlighting the mere fact that you hear more stories about how AI can place you in danger, than you hear stories of how AI can improve your safety. Therefore, this article will seek to address this stark imbalance in the literature, highlighting how artificial intelligence may just be the next revolutionary tool when talking about natural disasters and extreme weather events.

Since 1970 the number of natural disasters worldwide has quadrupled, and unfortunately 2018 does not appear to be an exception to this skywardtrend. In July we saw the worst floods the Indian state of Kerala has seen for a century, leaving 500 people dead and upwards of 1.5 million people displaced. Now I shall leave it to another article to suggest why this may be (ITS CLIMATE CHANGE!!), but with the frequency and the severity of said disasters set to continue long into the future, it only seems fitting that we adopt the aid of futuristic technologies to mitigate the damage and tragic loss of life. This is where AI comes in, it has the ability to: better predict the extent and areas of the damage, warn people earlier of the disaster, and better administer relief in cases where the disaster has already occurred.

For predicting areas where damage may occur, or as a tool for disaster planning, the ability AI has to collate and synthesise incredibly large data sets is invaluable. Take flooding for example: urbanisation has seen the amount of impermeable surfaces sky-rocket, increasing the amount of water run off which is unable to soak into the ground; consequently, leading to an increase in floods. An AI system developed by Microsoft however, has facilitated the creation of an incredibly accurate high-resolution map that can show objects as small as 3 feet. With such maps, ‘disaster planners’ are able to predict more accurately than ever before where rain water will travel and accumulate once it has fallen. Subsequently enabling them to predict what areas will be prone to flooding and make amendments as necessary. For example, they have been used to advise the Urban Forestry Administration in Washington DC as to where they should plant trees to avoid water accumulation, directly preventing the likelihood of floods. Using machine learning, the maps hope in the future to offer ‘live dashboards’, where automated alerts will notify the relevant authorities if an area is in risk of flood. Now many of you may be thinking that the maps will eventually go out of date, yet unsurprisingly AI is able to help in this region too. Microsoft has developed an algorithm allowing the map to identify changes on its own using aerial imagery, ensuring that they stay relevant and ensure the safety of the people within.

A caveat within disaster planning and issuing evacuation notices is that natural disasters such as earthquakes are notoriously hard to predict. However, by simulating earthquakes on a computer and gathering the data using machine learning, a team of geophysicists in America are now closer than ever to being able to forecast earthquakes.

Yet what if the disaster has already occurred? Well AI can also be applied to better know what types of relief will be most effective in certain areas. A company called One Concern takes scans of the structural integrity of buildings near to the seismic activity and analyses the demographics of the affected people. This information is then used by relief workers to determine what areas need what kind of assistance.

What is more, many disaster planners point to the key role social media may have in delivering relief. AI programs at the time of a disaster or extreme weather event could harness key information (such as marking yourself safe on Facebook) which could be used to understand the scale of the disaster and what areas need attention first.

Like all AI and like the Forbes article alludes to – its application here is not without its risks. Will people trust a company that collects social media in the wake of the Cambridge Analytica scandal? Will a wealthier demographic have its relief aid prioritised over that of another? It is unclear. But it is safe to say that recognising the potential of AI and its application to save lives is never going to be a bad thing, and coupled with checks and balances entrenched through legislature it has no reason to be feared. By better informing government officials, disaster planners and relief workers as to when a disaster is likely to occur, or to where aid is needed first, surely it should be celebrated – not demonised.