The development of artificial intelligence (AI) is concentrated in a few major tech hubs, creating a geographically uneven “superstar” dynamic. This concentration leaves many communities and demographic groups without the necessary resources for AI research and commercial scale-up. The issue of “compute” – the infrastructural elements driving digital systems – is central to this challenge, as AI models become increasingly complex and data-intensive.
- The National Artificial Intelligence Research Resource (NAIRR) aims to democratize AI research by addressing uneven access to AI-ready computing power, datasets, educational tools, and user support.
- The National Science Foundation (NSF) launched a pilot version of the NAIRR program with contributions from 12 agencies and 26 industry partners.
- The pilot’s first “allocation call” offered researchers access to advanced computing resources across six facilities.
The NAIRR pilot is an important step towards addressing the access challenge, but questions remain about its capacity to meaningfully advance geographic and demographic research inclusion in AI. The pilot faces limitations in providing state-of-the-art graphical processing units (GPUs) critical for training large language models, and recent budget cuts raise concerns about adequate funding.
Two key issues stand out: the extreme cost of access to resources needed for foundational AI research, and the persistent gaps in AI research and adoption skills at mid- and lower-tier research universities, particularly at Historically Black Colleges and Universities (HBCUs) and Minority Serving Institutions (MSIs).
To address these challenges, the NAIRR pilot should focus on expanding access to state-of-the-art computational resources, facilitating partnerships with universities and national labs, and investing in AI skills development. The pilot can play a crucial role in democratizing AI research by making resources more accessible to underserved academic communities and creating opportunities for low-resourced universities to build capacity.
