Thirst for AI

The world’s population is still rising and will reach 10 billion by 2050. This could lead to further exploitation of natural resources for agriculture; 70% of fresh water is used in that field currently. And as if that weren’t enough, climate change is making arid and desert areas encroach on more of the world. These three trends alone explain why the spotlight is always on our most precious asset. The data speak volumes. Of the 37 biggest aquifers in the world, 21 are shrinking. And it’s happening everywhere, be it in China, the United States or Italy, as a NASA study shows. 

The level of the Ganges basin continues to fall by 6.3 cm every year, as India’s ever-rising population draw too much on it for their farming. Due to global warming in China, the glacier on the Tibetan plateau is no longer a sustainable source of water for the Yangtze and Yellow rivers. Their waters have shrunk by 14% since the 1990s. The picture is disturbing. It points to an immediate future where less and less water is readily available. It’s not easy to see how this can happen in a planet whose surface is 70% water, and indeed where water has always been present and stable.

There are 1.386 million km³ of water on earth and it has been here in such stable quantities for billions of years, kept in place by the water cycle that brings evaporated water back down to earth. But of this tremendous quantity of blue gold, 97% is saltwater, which cannot be used for farming or human consumption. Another 2% is trapped in glaciers and polar ice caps. And even the remaining 1% is not entirely within our grasp. The fact is that humans can only use 0.5% of the water on earth. It comes from rivers, ponds, lakes, melting glaciers and underground aquifers, among other places.

This small percentage is being drawn on more and more, which is why water resources are running out, especially in times of drought, which are becoming more frequent. Water is being lapped up at a faster rate than it’s being poured back. And as if that still weren’t enough, the reserves are getting more and more polluted. Fertilizers used in agriculture ooze nitrates into reserves and can make the water in them unusable for drinking and irrigation. In brief, aquifers are drained, glaciers are melting and threatening their water supply, pollution is rising, and meanwhile the demand for water is only growing.

The result is that in recent years, some cities have found themselves up against real water crises. At the end of 2017 the people of Cape Town were almost left without tap water. The crisis was averted at the last minute with emergency rationing and a change of habits. The Australians managed to get through the drought that hit them from 1997 to 2009, but only by halving the amount of water they used. The Israelis have managed to tackle a constant shortage and actually become one of the best countries in the world at managing water resources.

They recycle and reuse waste water to satisfy 40% of the amount needed on all the country’s farms. One figure suffices to show just how much water we can save by using it more sparingly. In Italy, aqueducts tend to leak about 40% of that most precious resource. All they need to do is bring that number down sharply and there will be much less pressure on aquifers, which are are being drained in Italy too. This is one of the many fronts on which technology can help us stop wasting things we can’t afford to waste any more.

Artificial intelligence is the most promising candidate. Its capacity to unearth correlations in enormous rafts of data could really help us save our blue gold. For instance, a team of researchers at the University of Adelaide have come up with a system that uses deep learning (the algorithm that almost all artificial intelligence uses nowadays) to “make it much easier to spot cracks in aqueducts”. To succeed in this crucial effort, they have put sensors all along the water network which will monitor how the water flows, based on the pressure on the pipes.

When it analyses this data, the deep learning system will see when the pressure drops from a leak, thus locating it quickly and very accurately. This is just one of the many examples of how a data-driven approach to water management, using the extraordinary calculation skills of artificial intelligence, can make a real difference. Another example, again with aqueducts, is Fracta, a California start-up. It has come up with a system that spots where an aqueduct is very worn down and parts need replacing. Councils can use it to improve repairs and replacements, making them quicker and cheaper.

Fracta looks at the kind of soil the aqueduct sits on, the local topography, weather conditions in the area and a lot of other things, so it can estimate where the aqueduct is at imminent risk of leaks. Losses, waste and excessive consumption are also a feature of houses, but their inhabitants can spot them if they use smart water meters. The San Francisco company Valor Water uses artificial intelligence to analyse the anomalies picked up by these meters, and claims to be able to identify whether leaks or excessive consumption is the culprit. The company then advises its customers how they might waste less water.

But poor piping and excessive domestic consumption are only a fraction of the problem. Where action is needed is the 70% of water consumption occupied by agriculture. Waste needs reducing as much as possible, seeing that 60% of the time it’s due to inefficient use that could be corrected. Blue Tentacles is an Italian start-up whose precise AI-based system uses sensors to measure humidity, temperature, climate, weather forecasts and satellite data. The data it collects helps farmers irrigate their land as efficiently as possible, saving water and energy consumption.

With tools like this, they won’t need to water every single bit of land any more. Instead they can look at the data collected by sensors or drones and analysed by an artificial intelligence system, to see which patches or plants are in need of water, or which are drenched in too much water. Sensors buried in the ground measure how wet the soil is and send an analysis to an app on the farmer’s smartphone, which compares the data with weather forecasts and calculates how much water is needed for different crops, then automatically waters them. Digital technologies will also benefit the world of agriculture by greatly reducing pollutants, which harm water quality. 

A study by the European Parliament estimates that algorithms, drones and sensors linked up to a network –in other words, the Internet of Things– could let the agricultural sector cut down 85% on pesticides. Thanks to analysed data from sensors, farmers will know about problems from diseases and parasites well in advance. This will not only let them produce better-quality food, but also make their crops yield a lot more. So, to round off an absolutely classic virtuous circle, they will be producing more with the same amount of water.

Freelance journalist, writes about New Technologies, Politics and Society.