In 2018, Meta committed to minimizing its environmental footprint and is targeting net zero emissions for its value chain in 2030. However, it has plans to build eight data centers. To reduce the carbon emissions this one will generate, META's team, with the help of Lav Varshney and Nishant Garg from the University of Urbana-Champaign, designed a low-carbon concrete using generative machine learning algorithms that they tested at the Delkab, Illinois, facility. Concrete has been used for thousands of years to construct buildings and structures. Although it has evolved, cement is now one of its ingredients, but it is also the major source of its greenhouse gas emissions.

They may not be able to shout "Eureka!" like their human colleagues but AI/ML system have shown immense potential in the field of compound discovery -- whether that's sifting through reams of data to find new therapeutic compounds or imagining new recipes using the ingredients' flavor profiles. Now a team from Meta AI, working with researchers at the University of Illinois, Urbana-Champaign, have created an AI that can devise and refine formulas for increasingly high-strength, low-carbon concrete. Traditional methods for creating concrete, of which we produce billions of tons every year, are far from ecologically friendly. In fact, they generate an estimated 8 percent of the annual global carbon dioxide emission total. Advances have been made in recent years to reduce the concrete industry's carbon footprint (as well as in make the material more rugged, more resilient and even capable of charging EVs) but overall its production remains among the most carbon intensive in modern construction.

Porosity has been identified as the key indicator of the durability properties of concrete exposed to aggressive environments. This paper applies ensemble learning to predict porosity of high-performance concrete containing supplementary cementitious materials. The concrete samples utilized in this study are characterized by eight composition features including w/b ratio, binder content, fly ash, GGBS, superplasticizer, coarse/fine aggregate ratio, curing condition and curing days. The assembled database consists of 240 data records, featuring 74 unique concrete mixture designs. The proposed machine learning algorithms are trained on 180 observations (75%) chosen randomly from the data set and then tested on the remaining 60 observations (25%). The numerical experiments suggest that the regression tree ensembles can accurately predict the porosity of concrete from its mixture compositions. Gradient boosting trees generally outperforms random forests in terms of prediction accuracy. For random forests, the out-of-bag error based hyperparameter tuning strategy is found to be much more efficient than k-Fold Cross-Validation.