Illinois becomes integral in future of artificial intelligence
September 8, 2020
The University of Illinois will host two new artificial intelligence institutes funded by the National Science Foundation and the U.S. Department of Agriculture’s National Institute of Food and Agriculture. The NSF and USDA-NIFA are planning on investing more than $140 million in seven new artificial intelligence institutes across the country, two of which will be led here on campus. Investment into these new institutes will occur over a five-year period, driven by an initiative made by Congress and the White House.
The AI Institute for Future Agricultural Resilience, Management and Sustainability at the University will be funded by the USDA-NIFA and led by computer science professor Vikram Adve. AIFARMS will be tasked with developing AI-based techniques to impact the future of agriculture in two fundamental ways: magnifying the effectiveness of human labor and providing new methods of dealing with highly heterogeneous and diverse data across varying scales in both space and time. These two major goals have been divided into four broad research thrusts at the intersection of AI and agriculture: autonomous farming, livestock management, environmental resilience and soil health.
“AI has a number of important impacts on agriculture,” Adve said. “I should say that in these national AI institutes, there is a strong focus on both AI research and agriculture research. It’s not enough to look at applications of AI to solve agricultural problems. We also very strongly aim to improve the state of the art in AI itself.”
Adve further mentioned how agriculture today faces an enormous challenge in feeding a population that is growing worldwide, likely to increase by about 2 – 2.5 billion more people over the next 30 years. Technological advancements made during the Green Revolution in the 1950s-60s created severe impacts on the environment and on human health. Adve explained that we can not afford to incur those same kinds of impacts on the environment and human health in the next thirty years. “Conventional approaches to agriculture, including technological advances that have benefited agriculture in the past, are not enough to achieve our goals under very difficult constraints,” Adve said.
According to Adve, researchers at AIFARMS are extremely conscious of the adverse environmental impacts any new technology can have. “It’s a very high priority for us to be developing technology that has a beneficial effect (on the environment) rather than a harmful effect,” he said. Adve explained how a lot of negative impacts that present-day farming has on the environment are direct results of a limited number of farmers having to make large-scale decisions for huge plots of land. AI technology will be able to make guided decisions in localized areas to optimize farming and minimize the adverse environmental impacts we see today, such as water and soil contamination by the widespread use of herbicides.
The second AI Institute, funded by NSF, will be called the Molecule Maker Lab Institute and will be led by chemical and biomolecular engineering professor Hiumin Zhao. The MMLI has been established as a result of highly collaborative efforts made by a number of principal investigators in the institute as well as members of the University of Illinois campus. The focus of this institute shifts to molecular synthesis, the applications of which will be incredibly ubiquitous.
Functional molecules are used in all aspects of society. Most drugs, for example, are small molecule drugs that need to be synthesized. But not everyone can design certain molecules or chemicals, or if they can, the whole process can sometimes be very slow and time-consuming. “What we try to do at the MMLI is to develop AI tools to accelerate the discovery and manufacturing of those important molecules,” Zhao said. “Ultimately, I imagine all molecules can be made by automated instruments using AI tools. The goal is to produce billions of cell molecules.”
Zhao explained for a given molecule, there are many possible synthetic routes to produce it. However, it is unclear which one is the most economical or efficient. The AI tools at the MMLI will look at all synthesis strategies from the public domain and try to compare those different synthetic routes to propose the most efficient one. The chosen route will be implemented in the lab. “As you can imagine, it may not be successful the first time, so we learn from the trials and feed that information back into the development of the AI tools to increase accuracy,” Zhao said.
The MMLI will be targeting molecules with important applications. Some of these molecules might include antimalarial drugs or new phototactic materials that can harness solar energy more efficiently. Investigators at the MMLI are interested in producing chemicals or molecules that can be applicable in a variety of industries.
As new projects evolve and old ones remain ongoing, Zhao said that the institute will constantly evaluate all research endeavors to ensure that the developing AI tools are not used for harmful applications or that the chemical compounds cannot be optimized for use as chemical weaponry. However, the biggest challenge is that this field is so unexplored; there are no blueprints for navigating through this area of scientific discovery. “That’s why we want to engage the broader community to help develop new roadmaps for AI research with chemistry applications,” Zhao said.
Both AIFARMS and the MMLI have intentions of prioritizing community outreach. As successful molecular synthesis by AI tools will require the integration of many disciplines, the MMLI is already characterized by a diverse group of researchers. However, Zhao said that while the current team hails from a variety of universities boasting a variety of expertise, they would like to continue to engage external collaborators and expand their network. They would also like to engage undergraduate and younger students through research opportunities and outreach programs, make developing tools more accessible to all via tool kits and even host summer camps and competitions.
AIFARMS will be working with farmers and other professionals to ensure that developing research will actually be adopted and valued in the real world. The institute is also focusing on expanding education by developing new undergraduate and graduate degrees that intertwine computer science and agriculture. The goal is to develop a more skilled workforce and broaden participation in agriculture and STEM disciplines. They will be making efforts to bring in underrepresented groups, share teaching materials for use at other institutions such as Tuskegee University and work with Peoria school districts to facilitate advanced education in agriculture and STEM.
“One of our goals is not just to do the research ourselves, but also to set up a clearinghouse to coordinate external communities of people that are working on important problems for AI in agriculture,” Adve said. “We see our role there as laying the foundation for creating that kind of community.”
AIFARMS has plans to create an online platform filled with resources to help facilitate community engagement in solving problems for AI in agriculture. The hope is to eventually involve people on a global scale. “I think that’s a really important goal for us because it’s one way in which we think we can really magnify the impact of what we do,” Adve said.
Adve is the co-founder and co-director of the Center for Digital Agriculture and an affiliate of the Department of Electrical and Computer Engineering as well as the Information Trust Institute. Zhao is affiliated with the Departments of Chemistry, Biochemistry and Bioengineering as well as the Institute for Genomic Biology and the Center for Advanced Bioenergy & Bioproducts Innovation.
Shivali is a senior in AHS