We’ve all heard it. Our earth is changing.
The International Panel of Climate Change (2007, 2013) found in recent studies that “an increase of CO2 decreases the radiative cooling of the troposphere.” The use of fossil fuels worsen the natural greenhouse effect. This effect, also called global warming, warms the earth, makes ice glaciers melt, makes temperatures and sea levels rise and will cause more heatwaves. (NASA, n.d.)
“A change in climate can affect many related aspects of where and how people, plants and animals live, such as food production, availability and use of water, and health risks.” (Environmental Science Institute – University of Texas Austin, n.d.-b)
AI may be able to soften the blow of these effects. AI could help us monitor changes in the ocean that cause coral reefs to decrease. For example, The Ocean Data Alliance uses machine learning to help monitor shipping, ocean mining, fishing, coral reefs and their health or the outbreak of a marine disease. AI can keep track of the pollution levels in the ocean, track and follow marine litter and help to predict the spread of invasive species. (Cho, 2018b). This paper will discuss a scenario of an all-seeing AI and support it with research.
In 20 years the climate will have changed. Technology cannot keep that from happening as it is a natural course of events. But AI can help with the effects of climate change such as coral bleaching. In a few decades there will be an AI-system connected to stations all over the world, that tracks everything that happens on earth with support of big data. It can track and predict changes in temperature, changes in ground water, the movements of the tectonic plates and pollution levels in the ocean. For example, existing stations that record temperature data will be linked to the overseeing AI-system along with other data stations. When there’s a change in one of the subjects named above, the system will compare it to similar events that happened in the past and predict what will happen, when it will happen, how severe it will be and the best way to minimise damage.
For example, if we can oversee and/or predict the pollution levels in the ocean with the help of the all-seeing AI, we might be able to save coral reefs, save affected species and be one step ahead of climate change effects. There is a theory that climate change can cause more earthquakes, so besides tracking the changes of the earth and it’s climate, the system also gives us an opportunity to recognise (deep sea/underwater) earthquakes early on so we can take safety measures to mitigate disasters. By having an all-seeing AI we can track changes caused by climate change and do something about it.
“Artificial intelligence (AI) makes it possible for machines to learn from experience, adjust to new inputs and perform human-like tasks. Using these technologies, computers learn to accomplish specific tasks by processing large amounts of data and recognising patterns in the data.” (SAS, n.d.)
Coral reef bleaching is the loss of symbiotic algae that live inside coral due to environmental stresses (pollution or temperature changes). The loss of algae reduces photosynthetic pigment in corals, causing it to look white. (Environmental Science Institute – University of Texas Austin, n.d.-b & Van Oppen & Lough, 2018)
U.S Environmental Protection Agency, (2016b) found in a recent study that “if CO2 concentrations continue to rise at their current rate, the combination of climate warming and ocean acidification could slow coral growth by nearly 50%.” The ocean becomes more acidic as atmospheric CO2 dissolves in it. Coral reefs will decrease and die.
A recent publication of the Climate Council of Australia Limited, (2017, p.4) shows that “rising sea surface temperatures increase the frequency and severity of mass coral bleaching and reduce the opportunities for corals to recover.”
Dying coral reefs lead to “reduced habitat for marine organisms that depend upon the reef ecosystem, and fewer ecosystem goods and services, such as food, income and coastal protection, for dependent human communities” (National Oceanic And Atmospheric administration, 2017, p.1).
Recently The Nature Conservancy’s initiative teamed up with Microsoft’s AI for Earth and Esri to create an AI-powered web app. “The app can analyse geo-tagged underwater ocean images posted to Flickr. By matching the frequency and number of coral reef-related photos to other data, data scientists are able to quantify the value of coral reefs, kilometer by kilometer. (Microsoft. 2018, April 29)
“When pictures are uploaded to Google Photo, Google’s AI remembers the objects in the images. When you want to see a picture with a bicycle on it, it will display them. The AI has the ability to recognise the objects in the pictures. Facebook’s AI is able to recognise people in pictures and identify them out of approximately 3 billion people online. (Kevin Kelly, 2016, p.40)
However, a recent study (Athalye, Engstrom, Ilyas, & Kwok, 2017, p.1) shows that object recognition is not always right. “Neural network-based classifiers are vulnerable to adversarial examples”. “Adversarial examples are inputs to a neural network that result in an incorrect output from the network.” (Veerapaneni & Geng, 2018) The same publication shows the example of a picture of a panda, where the network is 58% certain that the picture shows a panda. Add some noise (an adversarial example), the network may think it’s a gibbon.
A good working recognition-function that is not vulnerable to adversarial examples, could be used in the all-seeing AI to monitor the health of different flora and fauna species without human intervention.
“In 2011, a 9.0 magnitude earthquake created a massive tsunami in Japan with waves reaching 40.5 metres.”
The disaster caused the worst nuclear meltdown since Chernobyl.
The city of Kawasaki remains at danger for future tsunami’s. It sits next to a fault line underneath the Nankai Trough, where earthquakes have produced damaging tsunamis. The city is also very close to dense populated city of Tokio. (Basu, M. 2017, 28 november)
Oceanographic and Atmospheric Administration (NOAA) utilize teleseismic and ocean bottom pressure measurements from DART buoys in the deep ocean.
The above mentioned organisations have set up an initiative, with help of AI and supercomputers, to study how to increase a tsunami’s profile accuracy, analyse flooding, evaluate in what rate actions could reduce tsunami damage and understand the characteristics of coastal tsunami behaviour. Right now there is no system that actually predicts tsunami’s and their profile. “Tsunami warning centers operated by the National Oceanographic and Atmospheric Administration use tele-seismic and ocean bottom pressure measurements from DART buoys in the deep ocean to pick up tsunami signals.” (Melgar & Bock, 2015)
The results of the Kawasaki coastal study will help to create effective mitigation measures for future earthquakes and tsunamis in other regions as well., (Fujitsu, 2017). The results of the study can also be used in the all-seeing AI to track, predict and analyse tsunami’s in other parts of the world.
Can climate change cause earthquakes? Volcanologist Bill McGuire’s answer is: yes.
When the earth emerged from an ice age, the climate changed, the temperature was rising and large ice sheets covering much of the planet retreated, causing the earth’s crust to ‘bounce’ back. This triggered earthquakes, tremors and volcanic activity. If global warming causes the ice sheets to melt, it could lead to more earthquakes just like 20,000 years ago. (Carbon Brief Staff. 2012, June 11 & McGuire, B. 2013)
Professors of the COMSATS University in Islamabad have build a probability model supported by AI to predict earthquakes using eight different identified seismic indicators. “Results indicate that the model can predict medium range earthquakes between 4.0 and 6.0 magnitudes.
During the testing period 2 of 4 earthquakes were predicted by this technique. (University of COMSATS Islamabad, 2014). This probability model supported by AI can be used in the all-seeing AI system to predict earthquakes in the future.”
There is a possibility of creating an AI that tracks everything happening in the world. Already there is a system in use that analyses the health of coral reefs, as mentioned before. If companies like Fujitsu create an AI supported system to prepare and protect coastal regions of Japan in case of tsunami’s, it can also be used in other areas of the world. A system that can predict earthquakes exists as well and could also be used in the all-seeing AI. The creation of a huge AI system that tracks the world might take several years. But it can be possible by combining the different existing systems, algorithms and technologies to create a bigger system that uses the data gathered by systems placed around the world. The more events happen, the smarter and better the AI gets. Just like IBM’s Watson. An AI will always improve.