The recent AI renaissance has led many to ask how this technology can help with one of the greatest threats to humanity: climate change. A new research paper written by some of the field's most famous thinkers aims to answer this question and gives a number of examples of how machine learning can help prevent human destruction.
The proposed uses vary, ranging from using AI and satellite images to better monitor deforestation, to developing new materials that can replace steel and cement (whose production accounts for nine percent of global greenhouse gas emissions).
But despite this variation, the paper (which we discovered via MIT Technology Review ) repeatedly returns to some broad placements . Prominent among these uses the machine view to monitor the environment. Using data analysis to find inefficiencies in the emissions industry; and use AI to model complex systems, such as the Earth's own climate, so that we can better prepare for future changes.
The authors of the paper ̵
"Only technology is not enough," writes paper writers, led by David Rolnick, a postdoctoral fellow at Pennsylvania University. "Techniques that would reduce climate change have been available years, but have largely not been adopted on a large scale by society. While we hope that ML will be useful in reducing the cost of climate action, humanity must also decide to act. "
Overall, the paper suggests 13 areas where machine learning could be used (of which we have chosen eight examples), which are categorized according to the timeframe of their potential impact, and whether the technique concerned is developed enough to reap some rewards. You can read the entire paper here, or browse our list below.
- Build better electrical system . Electricity systems are "redundant with data" but too little is done to take advantage of this information. Machine learning can help by forecasting electricity generation and demand so that suppliers can better integrate renewable resources into national networks and reduce waste. Google's British lab DeepMind has already shown this work with AI to predict the energy production of wind turbines.
- Monitoring agricultural emissions and deforestation . Greenhouse gases are not only emitted by engines and power plants – a lot comes from the destruction of trees, peat and other plant life that has captured coal through the photosynthesis process over millions of years. Deforestation and unsustainable farming cause this carbon to be released back into the atmosphere, but with the help of satellite images and AI, we can determine where this happens and protect these natural carbon sinks.
- Create new hydrocarbons. Papers authors note that nine per cent of all global greenhouse gas emissions come from the production of concrete and steel. Machine learning can help to reduce this figure by helping to develop alternative carbon dioxide emissions to these materials. AI helps researchers discover new materials by allowing them to model the properties and interactions of unprecedented chemical compounds.
- Assume extreme weather events . Many of the greatest effects of climate change over the coming decades will be driven by hugely complex systems, such as changes in cloud coverage and ice dynamics. This is exactly the type of problem that AI is good at digging in. Modeling these changes will help scientists predict extreme weather events, such as droughts and hurricanes, which in turn will help governments protect against their worst effects.
- Make transport more efficient . The transport sector accounts for a quarter of global energy-related carbon dioxide emissions, with two-thirds of this being generated by road users. As with the electrical system, machine learning can make this sector more efficient, reduce the number of lost journeys, increase vehicle efficiency, and move freight to alternative carbon emissions as a railroad. AI can also reduce car use by deploying shared autonomous vehicles, but the authors note that this technology is still not proven.
- Reduce waste energy from buildings . Energy consumed in buildings accounts for another one-quarter of global energy-related carbon dioxide emissions and presents some of the "lowest hanging fruit" for climate action. The buildings are long-lasting and are rarely retrofitted with new technology. Adding just a few smart sensors to monitor air temperature, water temperature and energy use can reduce energy consumption by 20 percent in a single building, and large-scale projects that monitor entire cities can have even greater impact.
- Geoengineer a more reflective soil . This use case is probably the most extreme and speculative for all those mentioned, but some researchers are hoping. If we can find ways to make clouds more reflective or create artificial clouds using aerosols, we can reflect more of the sun's heat back into space. It is a great though, and modeling of potential side effects of some systems is hugely important. AI can help with this, but the paper's writers note that there would still be significant "governance challenges" in the future.
- Give individuals tools to reduce their carbon footprint . According to the authors of the paper, it is a "common misconception that individuals cannot take meaningful action against climate change." But people need to know how they can help. Machine learning can help by calculating the individual's carbon footprint and marking small changes that they can make to reduce it – like using public transport more. buy less often; or reduce electricity consumption in their house. Setting up individual documents can create a large cumulative effect.