糖心视频

Research in education is struggling, but the good news is it can be revived, according to University of Kansas scholars.

In a new titled "The Death and Rebirth of Research in Education in the Age of AI: Problems and Promises," three KU education researchers address the biggest challenges facing their field and why now is a good time to consider not only the possibilities that artificial intelligence presents, but how and why the next generation of researchers can produce scholarship that has more influence on the field of education.

"Research in education has some fundamental issues it needs to deal with, and AI has exacerbated that in some ways. We've been doing this for a long time, and we haven't affected much or had the influence that we want to have or could have," said Rick Ginsberg, dean of KU's School of Education & Human Sciences and one of the article's authors.

In the article, the authors identify and analyze seven key problems facing education research:

1. Problems with peer review.
2. Quantification without contextual representation is tyranny.
3. The overblown research paradigm wars.
4. Overgeneralizing across contexts.
5. The negligence of individual diversity.
6. The typical vs. possible mindset.
7. The multiplicity and conflicting educational results.

In the case of problem No. 1, peer review is the standard model to ensure results are valid and based on sound science, but it leads to problems like reviewer burnout and delaying the editorial process so long that results are out of date before publication. The KU researchers also noted that scientists like Newton and Einstein published some of history's most important work without peer review.

In another example, the authors point out how problem No. 4, overgeneralizing, has led to overreliance on randomized control trials and assumptions that findings from one study of a group of individual students or educators will be typically true for all.

"Too often, research has prioritized what is typical and measurable over what is possible and meaningful," the authors wrote of problem No. 6, leading to a lack of imagination in education research.

The article, written by Yong Zhao, Foundation Distinguished Professor of Education; Neal Kingston, University Distinguished Professor of Educational Psychology; and Ginsberg, all of KU's School of Education & Human Sciences, was published in the journal

The authors write that while AI is not new, its recent improvements can present an opportunity at a time of stagnation in the field. While they noted it is not a replacement for researchers, it can allow for new ways of thinking, analyzing data and summarizing large bodies of data at a much faster than humans can. It also raises epistemological questions about what students should learn when machines can perform many cognitive tasks faster than people.

"It is inevitable that researchers come up against these challenges, and it's important to understand them," Kingston said. "People need to be guided to realize what is standing in the way of their research making a difference and improving things. AI is not a threat, and it's also not a panacea. But it can potentially help us improve."

Just as every individual is unique, every classroom is as well, and very few findings can be universalized to many or all educational settings. Given how that variability has held back success of research in some ways, the authors say instead of giving up on research, the time has come to reconsider how it can be transformed by rethinking educational aims, considering ethical, equitable and sociotechnical inquiry and thinking about issues such as distributed cognition, and how humans can use the technology to be co-learners with machines and democratize research by allowing students to be part of designing and guiding research.

That rethinking can be the foundation for a rebirth of educational research.

"We should treat AI as infrastructure, as another cognitive layer," Zhao said. "This is all to rethink and capture new ways to do research, and I think a lot of it is stuck in the past paradigm. Like anything new, there are risks, but we can find the best way to make it work. We wanted to summarize and introduce new possibilities, and hopefully this article will help guide people doing educational research going forward to think of new possibilities."