Responsible AI
Fostering collaboration, encouraging open source and advancing societal solutions.
Responsibly developing AI to benefit humanity
Meta has invested in the responsible development of AI for more than a decade, because we believe that AI has the potential to bring immense benefits to humanity. Open sourcing foundational AI technologies reduces barriers to entry; we share our data, models and learnings with the AI and broader community. Our AI investments allow us to play a role in addressing societal challenges, including those related to climate and society.
Designing data centers with AI in mind
Built with efficiency, flexibility and sustainability in mind, Meta’s next-generation data center will support our current products while enabling future generations of AI hardware. Our vision blends high-performance and power-efficient computing with a mix of custom hardware solutions specific to the needs of the environment.
Featuring denser racks to support large-scale AI clusters, along with future liquid-cooled AI hardware and network infrastructure, the design requires a smaller footprint to provide similar compute capacity as compared to previous data center designs. With an AI-optimized design, the next-generation data center improves delivery time and cost efficiency.
Open sourcing AI advancements
Meta shares our data, models and learnings with the AI and broader community. By open sourcing foundational AI technologies, we are reducing the barriers to entry for both technology companies with the computing infrastructure to train large models, and for the broader community of developers and organizations that stand to benefit from these innovations.
AI for climate change
Looking to AI to help identify solutions to the climate’s toughest challenges
-
Sustainable concrete
Concrete currently accounts for over 8% of carbon emissions worldwide. Accurate calculations for concrete can lead to reduced emissions, but this level of accuracy can be costly and time-consuming without AI. Meta developed an open source model that results in concrete mixtures with up to 40% lower-carbon concrete in collaboration with researchers at the University of Illinois at Urbana-Champaign.
-
Mapping the earth’s forests
Higher resolution forest data has the potential to dramatically improve the monitoring of conservation and restoration efforts globally. Leveraging our DINOv2 AI training algorithm and in collaboration with the World Resources Institute (WRI), we’ve launched an open source map of global tree canopy height at a 1-meter resolution. We hope this data can ultimately be applied to forest biomass and carbon stock monitoring.
-
Lowering cost barriers to carbon removal
Direct Air Capture (DAC) is a carbon removal solution that has the potential for large-scale impact, but is impeded by costs. DAC pulls CO2 out of ambient air through a recurring cycle facilitated by a sorbent or solvent, and the discovery of novel sorbents is one method to increase efficiency and reduce cost. Collaborating with Georgia Tech, Fundamental AI Research (FAIR) at Meta are working on OpenDAC, a research project aimed at reducing DAC costs. We released this dataset as well as state-of-the-art versions of our machine learning (ML) models trained on the dataset. As a result, more than 200 promising sorbents have been identified.
-
Identifying low cost catalysts
Carnegie Mellon University and AI at Meta are working together to identify more efficient and scalable ways to store renewable energy. Called Open Catalyst, the project aims to identify low cost catalysts that can drive the chemical reactions necessary to convert excess solar and wind energy into fuels that generate electricity.
Meta Llama Impact Grants
Our open source large language model, Llama, has been downloaded by hundreds of thousands of developers globally. To expand the opportunity to even more organizations across the globe, we’ve launched the Meta Llama Impact Grants program to identify and support the most compelling applications of Llama 2 for societal benefit.
2024 Sustainability Report
Learn more about our progress as we work to achieve net zero emissions across our value chain and become water positive in 2030.
