Intelligence through machine learning we begin to see more naturally in digital environments. Smart mails that practically write for us as you learn the algorithm, the frequency with which we use words, software that behaves according to our work and even closer, algorithms that show us what we like most in the entire spectrum that covers the internet .
MIT scientists have progress on how AI can help the planet cope with global warming, with some clearer ideas and others not so much, machine learning can directly contribute, for example, in regulating the use of electricity, if It is able to predict among other things the weather conditions in a specific place.
The report has been led by the postdoctoral expert of the University of Pennsylvania, David Rolnick. He has also received advice from several high-profile personalities, such as Google Brain co-founder and leading AI businessman and coach, Andrew Ng; the founder and CEO of DeepMind, Demis Hassabis; the general director of Microsoft Research, Jennifer Chayes; and the recent Turing Prize for his contributions to this field, Yoshua Bengio.
These are the 10 recommendations that best fit the implementation of machine learning. The full report can be viewed on this link.
If our dependence on renewable energy is going to be increasing, generation companies will need better ways to predict how much energy is needed, in real time and in the long term. There are already algorithms capable of forecasting energy demand, but they could be improved based on local weather and weather patterns or behavior in homes. Efforts to make the algorithms clearer could also help public service operators interpret their results and use them when programming when to start up renewable sources.
Scientists must develop materials that store, capture and use energy more efficiently, but the process of discovering new materials is often slow and without guarantees. Machine learning could speed things up by finding, designing and evaluating new chemical structures with the desired properties. This could, for example, help create solar fuels, which are capable of storing energy from sunlight, or identify more efficient carbon dioxide absorbers or structural materials whose manufacturing requires much less carbon. These new materials could replace steel and cement, whose production accounts for almost 10% of all global greenhouse gas emissions.
Shipping products worldwide is a complex and often very inefficient process. It requires the interaction of different shipment sizes, different types of transport and a changing network of origins and destinations. Machine learning could help you find ways to group as many shipments as possible to minimize the total number of trips. Such a system would also be more resistant to transport interruptions.
Electric vehicles are one of the key strategies to decarbonize transport. But its adoption faces several challenges in which machine learning could help. Algorithms can improve battery power management to increase the mileage of each charge and reduce "range anxiety or autonomy," for example. They can also model and predict the behavior of the added load to help network operators meet and manage their load.
Intelligent control systems can dramatically reduce the energy consumption of a building by analyzing weather forecasts, building occupancy and other environmental conditions. This analysis allows them to adjust the needs of heating, air conditioning, ventilation and lighting in an interior space. An intelligent building could also communicate directly with the network to reduce the amount of energy it uses if there is a shortage of low-carbon electricity supply at any given time.
Many regions of the world have little or no information about their energy consumption and their greenhouse gas emissions, which can be a major obstacle when designing and implementing effective mitigation strategies. Thanks to satellite images, artificial vision can extract traces and characteristics of buildings to power machine learning algorithms that can estimate energy consumption at a city level. The same techniques could also identify which buildings should be modernized to maximize their efficiency.
In the same way that machine learning can optimize shipping routes, it can also minimize inefficiencies and carbon emissions in the supply chains of the food, fashion and consumer goods industries. Better predictions of supply and demand could significantly reduce production and transport waste. On the other hand, specific recommendations on products with low carbon emissions could encourage more environmentally friendly consumption.
Much of modern agriculture is dominated by monoculture, which dedicates a lot of land to a single crop. This approach makes it easier for farmers to manage their fields with tractors and other automated basic tools, but also removes nutrients from the soil and reduces their productivity. As a result, many farmers rely heavily on nitrogen-based fertilizers, which can be converted to nitrous oxide, a greenhouse gas 300 times more potent than carbon dioxide. Smart robots could help farmers manage a mix of crops more effectively at scale, while algorithms could help farmers predict which crops they will plant, regenerating the quality of their land and reducing the need for fertilizers.
Deforestation contributes to approximately 10% of global greenhouse gas emissions, but its monitoring and prevention are usually a tedious manual process on the ground itself. Satellite images and artificial vision can automatically analyze the loss of the tree cover on a much larger scale, and soil sensors, combined with algorithms to detect chainsaw sounds, can help local authorities to stop illegal activity.
The AI techniques that advertisers have started using to target consumers can also be used to help us behave in a more environmentally friendly way. Consumers could receive personalized notices to promote their enrollment in energy saving programs, for example.
