The National Science Foundation (NSF) today announced the launch of a new public-private partnership called the National AI Research Resource (NAIRR) pilot program, aimed at democratizing access to AI research infrastructure and educational resources. The $60 million program brings together technology leaders like Nvidia, Microsoft, Google, Meta, and others to collaborate with academia and provide compute resources, datasets, educational materials and more to AI researchers across the country.
Public and Private Sectors Join Forces
The NAIRR pilot builds on President Biden’s National AI Research Resource Task Force report from last year, which recommended new models for public-private partnerships in AI. Per the NSF announcement, the program aims to “democratize access to AI through a federated system of research resources available to all US-based researchers.”
NAIRR will facilitate access to computing resources, data, educational tools and user support from some of the top names in the AI sphere:
- Nvidia – Providing access to DGX computing systems as well as the company’s GPU-optimized AI infrastructure
- Microsoft – Making Azure Cloud services available, including Azure compute, storage, databases, and machine learning capabilities
- Google – Contributing access to resources on Google Cloud
- Meta – Providing access to internal AI research training datasets
- Stanford – Leading development of teaching curriculum to onboard new users
In addition to enabling broad access to state-of-the-art research infrastructure, NAIRR aims to reach underrepresented communities with educational programs for using AI responsibly.
Empowering Responsible AI Innovation
While rapidly-advancing capabilities in AI hold tremendous promise, there are also risks relating to issues like bias, transparency, and data practices that need to be carefully managed. A core focus of NAIRR is education around developing and using AI responsibly.
“As artificial intelligence grows more powerful and more ubiquitous, it becomes imperative that it reflect the values we all share,” said Dr. Erwin Gianchandani, acting director of the NSF. “This means not only cultivating and diversifying the next generation of AI researchers, but also empowering communities to have a voice in how AI systems impact their lives.”
As part of the program’s commitment to responsible AI, NAIRR will require all users to complete trainings focused on trustworthy AI before getting access to advanced systems. Course topics will cover issues like:
- Mitigating unfair bias in AI systems
- Increasing transparency and explainability of models
- Protecting privacy through data security practices
- Monitoring for harmful applications of the technology
Table 1 shows some of the organizations contributing educational content and resources around responsible AI development as part of NAIRR:
|Stanford Institute for Human-Centered Artificial Intelligence (HAI)
|Leading development of onboarding curriculum on responsible AI practices
|Partnership on AI
|Providing trainings and collateral on trustworthy AI principles and frameworks
|Contributing resources on data ethics and avoiding AI harms
|University of Texas at Austin
|Developing guides on detecting and mitigating unfair bias in models
Empowering a diverse range of voices is critical for developing AI that works for everyone. In addition to building knowledge around responsible practices, NAIRR aims to increase participation in AI research from traditionally underrepresented groups like women, minorities, people with disabilities, and those based outside major tech hubs.
Accelerating US Leadership in AI
The launch of NAIRR signals a stepped-up national commitment to advancing AI research and solidifying US leadership relative to global competitors.
The program arrives alongside a sweeping update to US export controls aimed at limiting China’s access to certain AI chips and systems. By boosting domestic access to advanced research infrastructure, NAIRR aims to fuel US innovation in strategic technologies like AI while limiting technology transfer risks.
NSF Director Dr. Sethuraman Panchanathan positioned NAIRR as an effort to “future-proof” American leadership in science and engineering:
“We have an incredible opportunity to cement US leadership in artificial intelligence and protect our economic and national security by ensuring access to computing resources for researchers across the country.”
Industry leaders also spoke to the competitive angle of NAIRR inrelation to China and other nations pouring investment into AI:
“If America wants to continue leading in AI, we need to democratize access to AI infrastructure. NAIRR will play an important role by providing researchers access to advanced AI resources and educational tools,” said Jensen Huang, founder and CEO of Nvidia.
By facilitating cutting-edge research and establishing shared norms around responsible practices, NAIRR aims to drive US innovation in AI while steering the technology’s development in the public interest.
What Comes Next?
With the pilot phase now underway, NAIRR will spend the next six months onboarding users and iterating on the program:
- Onboarding Users – In the coming months, NAIRR will work closely with approved researchers to get them set up with access to computing resources, educational materials, and user support. Feedback from initial users will allow refinements before wider rollout.
- Developing Curriculum – Education partners like Stanford HAI will further build out trainings around responsible and ethical AI development to be required for all users.
- Exploring Funding Models – Stakeholders will explore sustainability models to support NAIRR’s national infrastructure and educational programs after initial pilot funding.
By mid-2024, NAIRR aims for full launch to provide AI access and education to thousands of researchers nationwide. Wider rollout would likely necessitate additional funding sources, whether from extended government grants, industry contributions, or a hybrid model.
If successful, NAIRR’s federated approach could provide a blueprint for democratizing access to emerging technologies beyond AI, like quantum computing. For now, execution in the pilot phase is critical for NAIRR to deliver on its vision of empowering responsible AI innovation across America’s research ecosystem.
To err is human, but AI does it too. Whilst factual data is used in the production of these articles, the content is written entirely by AI. Double check any facts you intend to rely on with another source.