Amazon, Google, Intel and Accenture will contribute toward a $160 million partnership for the next round of eight AI Research Institutes scheduled for creation in late 2021 by the National Science Foundation.

The announcement by the NSF marks the first time in which direct industry funding for the AI institutes will be received by the organization, which funded prior AI institutes on its own or with other governmental partners. Companies have participated in the NSF AI research institutes in the past with researchers, materials, content and more, but previously did not make direct monetary contributions.

“We’re partnering with industry to make available more resources to fund colleges and universities” to participate, Erwin Gianchandani, the deputy assistant director for the NSF’s directorate for computer and information science and engineering (CISE) division, told EnterpriseAI. “We’re bringing industry in where it wasn’t there before.”

The combined $160 million in new financial support from the NSF, other federal agencies and Amazon, Google, Intel and Accenture will be used to add new AI institutes to the NSF’s growing list of AI research programs, he said.

“And it’s not just about the funding – industry is also providing motivation, some of the researchers, questions the student teams might pursue, and they are in some cases providing access to data sets, access to compute resources, and other tools and testbeds,” said Gianchandani. “They are providing students in some cases with access to industry, which can be a proving ground and pathway to industry down the road.”

By partnering with industry, company researchers from Amazon, Google, Intel and Accenture can work side by side with college and university researchers and their students to work on AI projects within the institutes, he added.

The latest announced AI institutes are now soliciting applications for projects which will undergo rigorous review processes by the NSF to determine which college and university projects will be included in the AI institutes in 2021, said Gianchandani. The application deadline for the latest round of NSF AI institutes is Dec. 4. That process will only be conducted by the NSF and will not include the industry partners, he said. The NSF will continue to maintain total control over the institutes and the awards process as it has in the past, despite the new funding by industry, Gianchandani said firmly.

“We will make decisions and then hope to announce the awards next summer sometime,” he said.

The eight newly-announced NSF AI institutes that are accepting applications are:

• Human-AI Interaction and Collaboration: Enabling people to collaborate with intelligent machines toward common goals, based on fair and transparent principles that can be understood by users, and in increasingly complex interactions and contexts. Partial support provided by Amazon and Google.
• AI Institute for Advances in Optimization: Integrating perspectives from both classical constrained and unconstrained optimizations for solving previously “impossible” large-scale problems in planning, resource allocation, strategic reasoning, network and system design and optimization, hardware and software design and verification, and general combinatorial optimization and search. Partial support provided by the Intel Corporation.
• AI and Advanced Cyberinfrastructure: Integrating advanced cyberinfrastructure and AI as essential components of modern scientific discovery, to accelerate the ability to address pressing grand challenges in science and engineering and as a key enabler for advancing AI research and practice.
• Advances in AI and Computer and Network Systems: Exploring transformative ideas on how AI techniques can be used to advance computer and network systems, and how advances in computer and network systems can improve AI applications in the context of addressing long-standing as well as emerging challenges in computer and network systems.
• AI Institute in Dynamic Systems: Supporting research and education in fundamental AI and machine learning theory, algorithms, and use-inspired engineering and science for real-time sensing, learning, decision making and predictions that lead the way towards safe, reliable, efficient, and ethical data-enabled engineering and science systems.
• AI-Augmented Learning: Supporting research and development of AI-driven innovations to radically improve human learning and education writ large—in formal and informal settings, toward a grand challenge of “Education for All” through research on AI-supported learning systems to radically expand learning access to all Americans and in response to the rapidly changing landscape of jobs and work. Partial support provided by Accenture.
• AI to Advance Biology: Seeking bold AI-based advances and information infrastructure to push the frontiers of biology, increase our understanding of complex systems, and provide a theoretical basis for original research across the biological sciences.
• AI-Driven Innovation in Agriculture and the Food Systems: Advancing AI with application throughout agriculture and food production systems to overcome today’s major challenges in agriculture, build new multidisciplinary communities, and create the workforce needed for an AI-powered revolution in agriculture. This Institute theme is funded by USDA NIFA.

Amazon and Google will work with the upcoming NSF AI institute for Human-AI Interaction and Collaboration, while Intel will partner on the AI Institute for Advances in Optimization. Accenture will partner on the AI-Augmented Learning institute. Each of the industry partners will provide their expertise in those areas.

The future institutes will serve as nodes in a broader national AI research and education network and are expected to spawn new centers of AI leadership, according to the NSF.

The upcoming 2021 AI institutes will be the second round of AI institutes funded by the NSF. The winners of the first round, which submitted applications last October, were announced in August by the agency.

“We are hoping that by bringing together industry and colleges and government that we are catalyzing that connection for decades to come,” said Gianchandani. “You need more ideas to really understand and determine what are the problems.”

In the past, the NSF has worked with industry partners including joint solicitations with Intel, VMware and Amazon, but those projects did not include AI institutes.

“Most people in the academic community are very receptive to this because of the benefits,” said Gianchandani. “We’re doing this on our terms, and industry is working with us because they see the value propositions.”