Exploring the US Artificial Intelligence Implementation Plan for a National Research Resource
Summarising and Exploring the National Artificial Intelligence Research Resource Task Force’s 2023 Report
This article is a brief look at the report: “Strengthening and Democratizing the U.S. Artificial Intelligence Innovation Ecosystem: An Implementation Plan for a National Artificial Intelligence Research Resource”. The report was released to advise the President and Members of Congress. It counts 55 pages, yet with introductions and appendices it is closer to 100 pages.
As the report says in their introduction:
“We have only started to scratch the surface of what is possible, and cannot afford to miss out on seizing the opportunity for leveraging AI to serve the public good.”
So, let us see dive in.
1. Introduction
In this section I will try to cover the basic why, who and what.
Why this report? Who is it for?
“…provide AI researchers and students with significantly expanded access to computational resources, high-quality data, educational tools, and user support.”
In this way we could say it is intended to help researchers and students in the US mainly. However, to make this happen there is a large system of providers in education and infrastructure that needs alignment or coordination. It is outlined as an investment in foundational research to:
“…cultivate AI innovation and advance trustworthy AI.”
It does so while advocating for an inclusive approach in ‘engaging people from every zip code’.
What is this report meant to cover?
It is an implementation plan for a national cyberinfrastructure aimed at: “…overcoming the access divide, reaping the benefits of greater brainpower and more diverse perspectives and experiences applied to developing the future of AI technology and its role in our society.”
Who wrote this report?
It was written by the National AI Research Resource (NAIRR) Task Force.
Before we go further we will take a step back to explore what that means.
NAIRR you say?
2. Background
What is the National Artificial Intelligence Research Resource Task Force?
The Task Force has been composed of 12 technical experts: four from government, four from institutions of higher education, and four from private organisations.
Following the National AI Initiative Act of 2020, Congress directed the National Science Foundation (NSF), in consultation with the White House Office of Science and Technology Policy (OSTP).
Read more about the National AI Initiative Act of 2020 and other initiatives in the US at www.ai.gov:
OSTP and NSF formally launched the NAIRR Task Force in June 2021, appointing 12 leading experts equally representing academia, government, and private organizations
- Tess DeBlanc-Knowles (co-chair, beginning Aug. 2022), Senior Policy Advisor, National AI Initiative Office, White House Office of Science and Technology Policy
- Manish Parashar (co-chair, beginning Oct. 2021), Office Director of the Office of Advanced Cyberinfrastructure, National Science Foundation
- Lynne Parker (co-chair, June 2021 — August 2022), was Founding Director of the National AI Initiative Office, White House Office of Science and Technology Policy
- Erwin Gianchandani (co-chair, June 2021 — October 2021), Senior Advisor for Translation, Innovation, and Partnerships, National Science Foundation
- Daniela Braga, Founder & CEO of DefinedCrowd
- Mark E. Dean, Ph.D.
- Oren Etzioni, CEO, Allen Institute for AI
- Julia Lane, Professor, New York University; CEO, the Coleridge Initiative
- Fei-Fei Li, Sequoia Professor of Computer Science at Stanford University and Denning Co-Director of the Stanford Institute for Human-Centered AI (HAI)
- Andrew Moore, VP & General Manager, Google Cloud AI & Industry Solutions
- Michael L. Norman, Distinguished Professor, University of California, San Diego
- Dan Stanzione, Executive Director, Texas Advanced Computing Center/Associate Vice President for Research, The University of Texas at Austin
- Frederick H. Streitz, Chief Computational Scientist, Lawrence Livermore National Laboratory
- Elham Tabassi, Chief of Staff, Information Technology Laboratory, National Institute of Standards and Technology
3. What’s the Big Idea? Goals and Requirements
You read the typical schtick — AI will impact everything in society, transformative, consequences, etc — before describing what they want to do.
- Frontiers of AI is tied to access. There is a research resource divide that could skew the AI ecosystem.
- Platform/resource for AI. It is meant to bring together a range of aspects: “…computational resources, data, testbeds, algorithms, software, services, networks, and expertise.” AKA, almost Hugging Face for the government, right? 🧐
- Responsible AI. They argue their approach could help open up new opportunities also in areas such as: “…AI auditing, testing and evaluation, trustworthy AI, bias mitigation, and AI safety.” They want to set the conditions to develop systems inclusive by design.
- Competitiveness. They argue the CHIPS and Science Act of 2022 (fact sheet) reinforces the importance of: “…democratizing access to a national AI research cyberinfrastructure.”
For NAIRR they outline four measurable goals:
- Spur innovation.
- Increase diversity of talent.
- Improve capacity.
- Advance trustworthy AI.
The design should support data collection for assessment of these four goals.
Does it only include the US-based citizens?
“These users should be U.S.-based or affiliated with U.S. organizations, to include academic institutions, non-profit organizations, and startups or small businesses.”
As such, there is a possible interpretation to consider those working with or affiliated with US organisations.
How should it be governed? With a cooperative stewardship model.
“The NAIRR administration and governance should follow a cooperative stewardship model, whereby a single Federal agency serves as the administrative home for NAIRR operations and a Steering Committee comprising principals from Federal agencies with equities in AI research drives the strategic direction of the NAIRR.”
“A Program Management Office within the administrative home agency should provide funding and oversight for an independent Operating Entity that manages the day-to-day operations of the NAIRR.”
In short bullet points:
→ Cooperative stewardship.
→ Federal agency administrative home for operations.
→ Steering Committee with principals from Federal agencies.
→ Program Management Office in the administrative home agency.
→ Independent Operating Entity managing day-to-day operations.
How is this conceived? An integrated portal. Requirements? Should include:
- Conventional servers.
- Computing clusters.
- High-performance computing.
- Cloud computing.
- Access to edge computing resources.
- Testbeds for AI R&D.
- Tiered access structure.
- Ability to support a large number of users
- Ability to train resource-intensive AI models.
- Catalogs with search and discovery tools.
One provider? No no no.
“The Operating Entity should not itself operate the totality of the computer hardware that composes the NAIRR; instead, computing, along with data, testing, and training resources, should be delivered as services by partner resource providers selected through Federal agency or multi-agency funding opportunities.”
As such, it is supposed to be an operating entity together with a network of partner service providers.
4. Implementing NAIRR
How will it be implemented?
- In phase one, Congress should authorise and appropriate funds to establish the NAIRR.
- In phase two, the Operating Entity should establish its activities and oversee creation of the NAIRR portal and user interface, building in appropriate technical and policy controls.
- In phase three, the NAIRR should achieve initial operational capability and the Operating Entity should also formalise the policies, processes, and initial technical resources to be made available to AI researchers.
- In phase four, activities should transition from building out the NAIRR to establishing steady-state operations, as well as the planned evolution of NAIRR resources in response to user uptake and demand.
In regards to the first point, they have already made a: “Draft Legislative Language for NAIRR Authorization.” See Appendix J. As such, it should be somewhat easier to turn the proposal into a legislative text.
Parallell to this there is a proposed ‘pilot option’ (see p.40 in the report).
What is the proposed budget?
The Task Force estimates the budget for the NAIRR to be:
“…as $2.6 billion over an initial six-year period. The bulk of this investment ($2.25 billion) is to fund the resources to be made accessible via the NAIRR, through appropriations to multiple Federal agencies.”
How is this outlined for the resource providers?
“Resource providers should be brought online every two years with a six-year lifetime, so that a new $750 million investment is made every two years to ensure that the NAIRR resources remain state-of-the-art.”
How about the operating entity?
“The Operating Entity will require between $55 million and $65 million per year to support the coordination and management of NAIRR activities. An additional $5 million per year is budgeted for external evaluation of the Operating Entity and NAIRR performance.”
5. Conclusion
At this point we might be at the beginning of a shift in the American research community in AI both for researchers and students. Only time will tell. Regardless, it is a comprehensive piece of work that entails a possibility for the US to become more coordinated in their approach to AI broadly within research, especially in consideration of infrastructure as a central aspect of the development in the current field. If Congress decides to provide funds for the establishment of NAIRR investing anything close to the $2.6 billion outlined, then this will be a central national project in AI research on a global scale.
I will revisit this article and expand it over the coming week as I diver deeper into the report. Hope you stick with me and check back as I update this article and post some more reflections on this topic, especially thinking out loud about the possibilities for collaboration between Norway and the US on this NAIRR and the Norwegian Artificial Intelligence Cloud currently being developed here.
This is also part of my personal project #1000daysofAI and you are reading article 513. I am writing one new article about or related to artificial intelligence for 1000 days. The first 500 days I wrote an article every day, and now from 500 to 1000 I write at a different pace. On the one hand I release less articles now, on the other hand I am fortunate to be working with AI Policy, Governance, Ethics and International Partnerships full-time at NORA.ai.
