As faculty, researchers, and students turn to machine learning and AI across disciplines, best practices have emerged

5 developments around machine learning and AI in higher-ed research


As faculty, researchers, and students turn to machine learning and AI across disciplines, best practices have emerged

Using machine learning and AI in research is not limited to computer science and statistics–researchers in higher education are using it across a variety of disciplines, including life sciences and the humanities, according to new research.

The higher-ed community has been more proactive in exploring how machine learning and artificial intelligence (AI) can be used by researchers across higher education, a new EDUCAUSE report notes.

A partnership between EDUCAUSE and HP explores the types of machine learning and AI researchers use as they design and conduct their research. The partnership also looks into the methods and practices IT managers and departments use to determine the processes and infrastructure that best supports researchers at their institutions.

“The classic machine learning domains of computer science and statistics are continuing to push the boundaries of current knowledge, use, and application of machine learning and AI. But exciting new work is incorporating machine learning and AI into fields such as protein engineering, digital art, computational biology, civil engineering, and many more,” writes author Sean Burns, a corporate researcher at EDUCAUSE. “Institutions are also reporting more interest in and need for additional courses in machine learning for undergraduate students, while faculty are reporting higher application rates for master’s and PhD programs that involve machine learning.”

When it comes to IT needs and challenges, demand is on the rise for trained staff who can run and support machine learning technology, because maintaining this kind of technology requires very specific skills, according to the report. These skills include knowledge of how machine learning pipelines work and “how to help inexperienced researchers design and develop workflows for their research,” as well as the ability to “deploy, scale, and manage containers to help minimize virtualization footprints.”

Other IT needs and challenges around supporting machine learning and AI in higher-ed research include balancing tradeoffs of local versus cloud computing resources and making an effort to lower the barrier of entry to machine learning.

Despite these challenges, there are a number of encouraging IT best practices in this area: Focusing on funding and self-sustainment can help research continue; and IT teams can find proactive approaches to how they understand researchers’ needs and experiences. Included with these best practices are examples of them in action at different institutions.

“The growing role of machine learning and AI in higher education research is providing new challenges and opportunities for researchers and IT staff alike,” says Burns. “Working to support the ingenuity of faculty and the growth of machine learning in science and research is something every institution can strive for.”

Other key findings include:

1. The use of machine learning is not limited to computer science and statistics. Researchers are beginning to explore how machine learning can improve research across many different disciplines in engineering, life sciences, and the humanities as they gain access to new and larger datasets.

2. Not all machine learning users are created equal—they have different technical ability levels. As the number of disciplines and faculty engaging in machine learning expands, many more researchers can be found who are just starting to learn how to use machine learning in their field of study.

3. Building communication lines between IT and researchers is key for effective machine learning support. Both IT and researchers yield better outcomes with fewer resources when they communicate early and often about needs and goals.

4. Machine learning is costly and requires substantial support. Institutions often lack the internal resources for supporting machine learning. As a result, funding through grants and other external sources is often necessary for both researchers and IT to obtain modern machine learning hardware.

5. Institutions are working to lower the barrier of entry to machine learning. As more and more researchers from backgrounds other than computer science are exploring machine learning, institutions are developing training and new resources to help make machine learning more available for the uninitiated.

Sign up for our newsletter

Newsletter: Innovations in K12 Education
By submitting your information, you agree to our Terms & Conditions and Privacy Policy.

Laura Ascione

Sign up for our newsletter

Newsletter: Innovations in K12 Education
By submitting your information, you agree to our Terms & Conditions and Privacy Policy.