In the era of generative artificial intelligence (AI), data is king. The large language models on which AI tools are built rely on vast amounts of data, and training and managing AI tools requires powerful (and often expensive) computing power, requiring major technology providers are far ahead of other companies in advancing AI. It creates a potential threat to everyone else.
As generative artificial intelligence transforms the way we work and live, a group of higher education leaders is joining forces to ensure that universities, especially those with limited resources, are not left behind.
The American Council on Education (ACE) is leading the effort to establish a global data consortium. Through the Global Data Consortium, participating universities, nonprofit organizations, and others around the world contribute institution-level datasets about their learners, creating a wealth of information. Researchers, academic leaders, and in some cases, businesses will also use the consortium to build his AI products and deepen our understanding of what works (and what doesn't) in student life. can.
“The way higher education shapes its future is by owning the data,” said Ted Mitchell, president of ACE.
Officials at ACE and its partners believe that the technology that will be used to build the network, an artificial construct designed to mimic the statistical properties of real-world data without compromising individual privacy. (includes the use of “synthetic datasets'') and “''. “Meshing'') will allow participants to glean insights from the data without actually accessing the underlying student information, remaining firmly in the control of the participating universities.
Paul LeBlanc, who is stepping down as president of Southern New Hampshire University and is focusing his next career on artificial intelligence, said enough major institutions, governments and others have committed to the project that there are already “3,500 10,000,000 pieces of data will be saved.” We support our students from the day they turn it on. ”
“We seem to be getting over the idea that your data is yours. [institution’s] It's a competitive advantage,” LeBlanc said. “Right now, we have to be very reactive: What is Open AI creating and how can we take advantage of it?”
He said working together is the best way for higher education to “own our future” rather than letting Google, Meta, Amazon and other major tech powers dominate the world of AI. He added that it was an opportunity.
Scott Durand, a former Southern New Hampshire official who oversees ACE's project, said this is especially important for “smaller, under-resourced institutions that are going to have the hardest time in the AI game.” Ta.
Major universities like Arizona State University and the University of Michigan have agreements to collaborate with Open AI, and others are creating their own AI tools, but all have the resources to do it well. It doesn't mean you have it.
Universities without significant institutional research offices will gain insight into how their student bodies compare to learners at other universities and benefit from AI tools jointly developed by consortium members. can do. Durand also said the consortium envisions providing technical assistance to universities in need.
University officials briefed on the project expressed excitement but acknowledged potential hurdles.
Pennsylvania Higher Education System Chancellor Dan Greenstein saw the potential to “accelerate the pace of innovation” through this type of data sharing and collaboration.
“We may be able to avoid people going into dead ends or replicating work that others have already done,” he said.
However, he stressed that participating institutions must meet high standards to ensure the quality of the data they provide and must agree to abide by the limits the consortium sets on how the data is used. “Members have certain obligations,” he said.
“Data will be shared on a solution-oriented basis,” says a technical white paper on the project co-authored by George Siemens, professor and director of the Center for Change and Complexity in Learning at the University of South Australia. “This means that data is shared that is relevant to the specific functionality that a consortium project or member wants to achieve. This ensures that the data made available is not shared for the sake of sharing, but is based on clear needs. We will be able to respond reliably.”
The consortium is in talks with major foundations for initial funding, and the effort could eventually be funded through membership fees, data access fees, or other forms of funding.
Institutions interested in learning more can sign up here to stay up to date.