Details

Autor(en) / Beteiligte

• Korolev, Vadim
• Mitrofanov, Artem

Titel

The carbon footprint of predicting CO2 storage capacity in metal-organic frameworks within neural networks

Ist Teil von

• iScience, 2024-05, Vol.27 (5), p.109644-109644, Article 109644

Ort / Verlag

Elsevier Inc

Erscheinungsjahr

2024

Link zum Volltext

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• While artificial intelligence drives remarkable progress in natural sciences, its broader societal implications are mostly disregarded. In this study, we evaluate environmental impacts of deep learning in materials science through extensive benchmarking. In particular, a set of diverse neural networks is trained for a given supervised learning task to assess greenhouse gas (GHG) emissions during training and inference phases. A chronological perspective showed diminishing returns, manifesting themselves as a 28% decrease in mean absolute error and nearly a 15,000% increase in the carbon footprint of model training in 2016–2022. By means of up-to-date graphics processing units, it is possible to partially offset the immense growth of GHG emissions. Nonetheless, the practice of employing energy-efficient hardware is overlooked by the materials informatics community, as follows from a literature analysis in the field. On the basis of our findings, we encourage researchers to report GHG emissions together with standard performance metrics. [Display omitted] •A diverse set of NNs for predicting CO2 capacity in MOFs was trained•A disproportionate increase in model complexity and carbon footprint was noted•Environmental effects of AI in materials science need more quantification Global carbon cycle; Applied sciences; Algorithms.