Artificial Intelligence in all its guises has captured much of the conversation in HPC and general computing today. The White House, DARPA, IARPA, and Department of Energy all have issued strategies or undertaken programs intended to foster AI development and use. Yesterday, the Computing Community Consortium (CCC) weighed in with a 100-plus page draft report – A 20-Year Community Roadmap for Artificial Intelligence Research in the US – and CCC is seeking comment around its concepts and recommendations.
The CCC, of course, is a body formed to “define the large-scale infrastructure needs of the computing research community” that was created in response to a National Science Foundation (NSF) solicitation in 2006. In turn, the CCC is part of the Computing Research Association (CRA) founded in 1972 and encompassing academia, industry, and government; its proposals, among other things, help inform NSF activities and federal computing priorities.
As noted on the CCC website, “The CCC Council meets three times every calendar year, including at least one meeting in Washington, D.C., and has biweekly conference calls between these meetings. Also, the CCC leadership has biweekly conference calls with the leadership of NSF’s Directorate for Computer and Information Science and Engineering (CISE).”
CCC began work on the new AI roadmap last fall, held three workshops and a ‘Town Hall’ meeting in 2019, and yesterday issued a blog calling for comment on its roadmap. Comments are due by May 28, 2019.
Honestly, parsing such a large document is best done by directly reading it and CCC has packed its AI roadmap with all manner of observation and suggestion. Here are its major recommendations excerpted from the bog:
I – Create and Operate a National AI Infrastructure to serve academia, industry, andgovernment through four interlocking capabilities:
a) Open AI platforms and resources: a vast interlinked distributed collection of “AI-ready” resources (curated high-quality datasets, software libraries, knowledge repositories, instrumented homes and hospitals, robotics environments, cloud-scale computing services, etc.) contributed by and available to the academic research community, as well as to industry and government. Recent major innovations from companies demonstrate that AI breakthroughs require large-scale hardware investments and open-source software infrastructures, both of which require substantial ongoing investments.
b) Sustained community-driven AI challenges: organizational structures that coordinate the formulation of grand-challenge problems by AI and domain experts to drive research in key areas, building upon—and adding to—the shared resources in the Open AI Platforms and Facilities.
c) National AI Research Centers:physical and virtual facilities that bring together Faculty Fellows from a range of academic institutions and Industry Fellows from industry and government in multi-year funded projects focused on pivotal areas of long-term AI research.
d) Mission-Driven AI Laboratories:living laboratories that provide sustained infrastructure, facilities, and human resources to support the Open AI Platforms and the AI Challenges, and work closely with the National AI Research Centers to integrate results to address critical AI challenges in vertical sectors of public interest such as health, education, policy, ethics, and science.
II – Re-conceptualize and Train an All-Encompassing AI Workforce, building upon the elements of the National AI Infrastructure listed above to create:
a) Development of AI Curricula at All Levels: guidelines should be developed for curricula that encourage early and ongoing interest in and understanding of AI, beginning in K-12 and extending through graduate courses and professional programs.
b) Recruitment and Retention Programs for Advanced AI Degrees: including grants for talented students to obtain advanced graduate degrees, retention programs for doctoral-level researchers, and additional resources to support and enfranchise AI teaching faculty.
c) Engaging Underrepresented and Underprivileged Groups: programs to bring the best talent into the AI research effort.
d) Incentivizing Emerging Interdisciplinary AI Areas: initiatives to encourage students and the research community to work in interdisciplinary AI studies—e.g., AI-related policy and law, AI safety engineering, as well as analysis of the impact of AI on society—will ensure a workforce and a research ecosystem that understands the full context for AI solutions.
e) Training Highly Skilled AI Engineers and Technicians, to support and build upon the Open AI Platform to grow the AI pipeline through community colleges, workforce retraining programs, certificate programs, and online degrees.
III – Core Programs for AI Research are critical. These new resources and initiatives cannot come at the expense of existing programs for funding theoretical and applied AI. These core programs—which provide well-established, broad-based support for research progress, for training young researchers, for integrating AI research and education, and for nucleating novel interdisciplinary collaborations—are critical complements to the broader initiatives described in this Roadmap, and they too will require expanded support.
As you can see, there’s a lot here and that includes calling for increased spending in the context of a global race for AI. The report declares, “U.S. leadership in AI is at risk without significant, strategic investments, new models for infrastructure and resources, and attention to the education and training pipeline. Other major industrialized countries are already embarking on substantial AI research programs.
- “The EU has announced funding of 20B Euros for AI17 and is currently evaluating proposals for decadal-long 1B Eur0 science projects, one of them in the area of AI assistants. Germany and France have allocated 3B and 1.5B Euros to AI, respectively. The UK has pledged an investment of 1B Pounds in AI, together with dedicated funding for 1,000 PhDs and 8,000 specialized teachers in AI, and has repurposed its flagship Turing Institutes into major data-driven AI research centers.
- China has announced that it will invest billions in AI over the next five years, creating at least four $50M/year AI Centers and a $1B/year National AI Research laboratory with thousands of AI researchers and engineers, and committing to training 500 instructors and 5,000 students at major universities…”
The scope of these efforts, argues CCC, “are in line with major U.S. research investments in the past, such as the LIGO project ($1.1B), the Human Genome project ($2.7B), and the Apollo program ($144B), all of which not only led to major scientific advances, but also produced significant economic and societal benefits.”
The recommended national Pivot AI Research Centers (PAIRCs) intended to create unique and stable environments for large multi-disciplinary teams devoted to long-term AI research aren’t cheap. The report says, “Each PAIRC would be funded in the range of $100M/year for at least 10 years. With this level of funding, a PAIRC would be able to support an ecosystem of roughly 100 full-time faculty (in AI and other relevant disciplines), 50 visiting fellows (faculty and industry), 200 AI engineers, and 500 students (graduate and undergraduate), and sufficient computing and infrastructure support.”
By way of analogy, the report notes, “There are a few examples of AI research centers that have long-term funding. The University of Maryland’s Center for the Study of Language (CASL), founded in 2003 as a DoD- sponsored University Affiliated Research Center (UARC) funded by the National Security Agency, includes about 60 researchers and 70 visitors from academia and industry focused on natural language research with a defense focus.”
How the report translates into NSF or further US-funded AI activities, of course, remains to be seen.
Link to draft AI Roadmap: https://cra.org/ccc/wp-content/uploads/sites/2/2019/05/AIRoadmapDraftforCommunityMay2019.pdf
Link to comment form: https://computingresearch.wufoo.com/forms/s15u6ssf15mvnlg/
Figures all taken from the draft report