Meta Platforms Inc, which was formerly known as Facebook, has been seen to be using artificial intelligence to address climate change problems and develop elegant engineering solutions, with the aim of increasing the efficiency of industrial systems with technologies of machine learning.

There is also a partnership between Meta AI and the Department of Chemical Engineering at Carnegie Mellon University, with which the Open Catalyst Project represents one of these initiatives, which is based on helping bring together various artificial intelligence researchers to design new ML models with the purpose of predicting new chemical reactions for energy storage.

For example, the idea is to build energy storage, which can be batteries that manage to absorb excess energy so that it can be delivered outside of peak hours, since renewable energy is generally solar energy and wind energy, which depend on the availability of the sun and the wind to generate energy respectively, however, at times when there is neither sun nor wind, energy generation decreases for a period of time, therefore the idea of creating a Energy storage is a great idea. However, Larry Zitnick, Principal Research Scientist at the Open Catalyst Project mentioned that “The problem is that batteries don’t scale really well with storage,” but still you can find a way that can help. to store energy “So we need to find a way to store energy that actually scales. That’s where the Open Catalyst Project actually comes in.”

Project

Open Catalyst has created the largest materials training dataset, together with the research team, in order to have better renewable energy storage in the world. This project provides data for the discovery of chemical catalysts for the purpose of building batteries much cheaper and that can be scalable for renewable energy networks.

The process of this research can be long and slow, since according to Zitnick, there will most likely be millions of different material combinations that could be tested in laboratories, and these materials could be tested at a rate of only about a thousand per cent. year. Still, with over 8 million data points and 40,000 unique simulations across a variety of Open Catalyst materials, it can help jumpstart and better research.

In addition, Zitnick mentioned that “brute force” simulations in seconds are done with computational computing on the system, which can help in a big way, because with other systems it could take researchers days to find workable optimizations.

AI use

A senior member of Meta, Mike Schroepfer, has mentioned that Meta’s global operations are supported by 100% renewable energy, however, the efficiency of these energies is crucial and the company is exploring green AI models, therefore it should be Keep in mind that the use of AI with the purpose of managing to face the climate crisis is precisely with the efficiency of its own infrastructure and energy needs.

It is very important and necessary to train AI models against ever-growing and complex data sets, as optimizing AI models, especially powerfully built ones, on a large scale can be particularly problematic when running. 

It was found in an experiment where researchers were able to identify optimizations which reduced the infrastructure resources used for language translation by 800 times. Therefore, for natural language processing, translation and even the use of AI in the platform, this potential at this level of algorithmic optimizations and performance gains manages to have too great an impact on the emissions caused by the use of the AI.

Therefore, there are high expectations about the impact that AI will have on climate and sustainability, in this work that is being carried out with Meta and its researchers.