AMHERST — Artificial intelligence could be key in leading us toward a greener future. At the University of Massachusetts Amherst, researchers were just awarded $12 million by the National Science foundation to develop a new field that aims to use artificial intelligence to shrink society’s carbon footprint on a large scale.

Researchers say their findings could be used to automate decarbonization across sectors of the economy, making huge steps to reduce carbon emissions possible. Rather than striving for energy efficiency as a mode of emissions reduction, computational decarbonization, or CoDec, will focus on the emissions themselves.

“Energy efficiency has been a focus in research, policy, and other areas for many years now. It allows you to reduce your energy consumption, but at the end of the day, it doesn’t always significantly reduce your carbon emissions,” said Prashant Shenoy, distinguished professor and associate dean for the Manning College of Information and Computer Sciences (CICS) and principal investigator for this research initiative. “We’re seeing the effects of climate change with changing weather patterns and more, so now we’re trying to reduce our carbon footprint by making infrastructure carbon-efficient.”

According to Shenoy, energy efficiency tends to focus on the implementation of efficient energy sources such as wind or solar, while carbon efficiency directs focus to the timing of energy use to minimize the amount of work being done with emissions-heavy energy sources. For example, Shenoy said that on sunny days, those with solar power have access to greener energy. But when it gets cloudy, their energy use becomes “browner” as backup forms of energy are employed. CoDec hopes to use artificial intelligence to let individuals know the best times to use the most energy, based on when they have the most green energy available.

“To achieve carbon efficiency, you need to track when energy is greener and do certain kinds of work, like maybe your laundry, during those times,” Shenoy said.

Using a sense-optimize-reduce framework, commonly used in cyber-physical systems, CoDec seeks to define carbon footprints in different areas, create computational methods and algorithms to optimize the amount of carbon emitted within a given system, and develop software that allows solutions to be implemented in the real world. CoDec will examine carbon efficiency across computing, electricity, buildings and transportation infrastructure.

“The different infrastructures we’re looking at in this project all have different dimensions of flexibility in terms of time and space,” says David Irwin, a professor of electrical and computer engineering and one of the UMass Amherst contributors on this project.

“For example, the heating in a building can’t be shifted in space — the building can’t move. However, computing is uniquely flexible across time, space and performance. You can move running a computing job or serving a web page from New York to California. And you can do it very quickly. But all of these infrastructure systems have some flexibility that we’ll be looking to exploit.”

While CoDec will use artificial intelligence to seek carbon-efficient infrastructure solutions, researchers acknowledge that the technology is very energy-consuming in itself. Throughout the research process, the team at UMass will also be seeking ways to make AI systems more carbon-efficient.

One of the main elements of this new research is looking at the interconnected nature of infrastructure systems that makes decarbonization a tricky task. Shenoy explained that this issue is called the interdependency gap.

“One of the technical focus areas of this research is looking at decarbonization more holistically,” said Shenoy. “When you try to reduce a carbon footprint in one sector, it may actually increase in another sector. … We want to tackle the problem by taking a good look at these interdependencies.”

Shenoy explained these interdependencies with the example of working from home, which reduces one’s carbon emissions from transportation, but shifts much of their carbon burden to their home energy use.

CoDec research also seeks to establish long-term computing techniques, as computing is typically concerned with time frames measured in milliseconds, but carbon optimization — emitting the lowest possible amount of carbon in a given sector — must be addressed over the span of months or even years.

To address these multifaceted research questions, the CoDec project will develop theoretical foundations of computational decarbonization by combining safety-critical optimization and data-driven learning techniques. The researchers will also develop new courses in this emerging field and work with a group of industry partners to transition decarbonization research to real-worldpractice.

“I’m really excited about all of the research we’ll be doing with this interdisciplinary team. We have researchers from backgrounds in engineering, computer science, economics, and other areas,” said Shenoy. “We all want to bring ideas from this research into the classroom and get students excited about these issues and their potential solutions.”

The CoDec initiative is one of three so-called Expeditions projects receiving a portion of $36 million from the NSF, which aim to “[create] transformative technologies, methodologies and infrastructure that can be adopted by the broader research community, industry or society at large,” according to the NSF. Other institutions involved in Expeditions projects are Carnegie Mellon University, Massachusetts Institute of Technology, the University of Chicago, the University of California Los Angeles and the University of Wisconsin.