Currently, artificial intelligence (AI) is considered the most important technology in the 4.0 Industrial Revolution, which can create changes in the global order.

Some great powers such as the US, Japan, China and some EU countries have issued and built long-term strategies to promote the development of AI technology.

To consolidate world leadership in AI technology as well as more strongly promote research and development (R&D) activities, build and apply reliable AI systems to serve development benefits socio-economic, the US Congress passed the "National AI Initiative Act of 2020". Implementing the above law, the White House Office of Science and Technology Policy (OSTP) has coordinated with the National Science Foundation (NSF) to establish a "Task Force" on "National AI Research Resources" (NAIRR) at the end of June 2021.

NAIRR is comprised of 12 leading U.S. AI experts, representing government agencies as well as academic institutions and private organizations, and functions as an “Intergovernmental Advisory Committee” with its mission to research and identify a roadmap for building a national AI research infrastructure that can share and expand access to resources needed for US AI R&D activities.

The basis for forming a roadmap for building a national AI research infrastructure

Recently, on January 26, 2023, the Office of OSTP released the final report of NAIRR "Consolidating and democratizing the US AI innovation ecosystem: Implementation plan exploiting national AI research resources". This is also the roadmap to build a national research infrastructure for AI R&D to maximize the benefits of AI, expected to launch from 2023. At the same time, the roadmap also ensures that the systems developed by AI-driven does not harm people's rights or freedoms and increases America's global competitiveness in this important technology sector.

Much of today's AI research in US relies on access to very large volumes of data and advanced computing power. Such research conditions often cause obstacles and difficulties for those working at technology companies, research facilities and universities that are not equipped with modern infrastructure. It is this division of access that indirectly limits the range of ideas incorporated into AI innovation, contributing to the biases and other systemic inequities found in today's AI R&D systems in US.

Therefore, to overcome this inequality, US has established a roadmap to build national AI research resources to create a shared, nationwide AI research infrastructure that will provide grants researchers, teachers, and students in the field of AI significantly expanded access to high-performance computing resources, high-quality data, and stronger educational and promotional tools R&D activities on AI.

Recommendations for a roadmap to build a national AI research infrastructure

NAIRR’s report helps expand the range of researchers involved in AI R&D, developing and diversifying the approaches and applications of AI technology. Therefore, the report will help solve national problems, building a roadmap to transform the US national AI research ecosystem by strengthening and democratizing AI R&D in basic research. publication, applied research and commercialization of research results.

According to the report, US will focus on incorporating diversity and access to AI R&D resources to create new technologies as well as efforts to democratize access into emerging technology. That ensures equitable use of machine learning systems, connecting the US. AI research community with big data, supercomputing for complex computational processing, and efficient testing capacity. results to advance AI research through a user-friendly interface as well as user support and related training.

In particular, the main content explains and makes feasible recommendations on a number of core issues about NAIRR, specifically as follows:

- NAIRR should be established with four measurable goals: (1) promote AI innovation, (2) increase AI talent diversity, (3) enhance AI research capacity and (4) advance trustworthy AI systems;

- Management and administration of NAIRR should follow a collaborative management model, whereby only one federal agency will serve as the lead agency for the governance of NAIRR activities, namely NSF and a “Board of steering” includes the heads of federal agencies that fund AI research and decide on the strategic direction of NAIRR;

- NAIRR will provide user access to a research infrastructure that harmonizes data and computational resources, test facilities, software and testing tools. as well as user support services through an integrated portal;

- NAIRR must be widely accessible to users, expand the range of researchers involved in AI R&D, provide a usable platform for educational activities, and reduce barriers to participation in the system. AI research ecology;

- NAIRR should establish standards of accountability in AI research through the design and implementation of AI governance processes;

- NAIRR should implement security protections according to established guidelines;

- NAIRR will have huge and far-reaching impacts, allowing AI researchers to solve problems ranging from routine tasks to global challenges.

- NAIRR - construction implementation plan is expected to be implemented in 04 phases, starting immediately after the publication of this report.

- The total estimated cost to implement the NAIRR construction roadmap is about 2.6 billion USD, which needs to be invested from the federal budget over a period of 6 years and according to each implementation stage.

Conclusion

Overall, US community of AI scientists appreciated the announcement of this deployment plan. Because NAIRR will help reduce the barriers to entry into AI R&D and education, while also making it easier and less expensive for researchers to access core AI research tools.

In particular, NAIRR development roadmap and management model can be researched and applied in Vietnam to contribute to the implementation of the goals stated in the National Strategy on Research, Development and Application. AI to 2030, issued in Decision No. 127/QD-TTg dated January 26, 2021 of the Prime Minister. This can also be a source of information for ministries and branches to refer to, research, build and complete policies to develop research infrastructure for Vietnam's AI technology and develop parallel cooperation plans. on AI technology with the United States in the coming years./.

Source: SATI