The Promise and Risks of A.I. in Education

The latest push for schools to keep up with sweeping, societywide technological changes is coming at the same time that American students are struggling. Test scores have been declining for over a decade along with basic numeracy and literacy. There are lots of explanations for why that’s happened, though the two most obvious culprits are the ubiquity of screens and the decline of accountability measures. Covid made it clear that a kind of magic happens in a physical classroom through human connection that cannot simply be replaced by apps. Presumably A.I., even when carefully introduced and overseen by trained educators, risks killing some of that magic.

As a result of these deficits, fewer students appear to be ready for the rigor of college. According to a report from the U.S. Chamber of Commerce and the College Board, 80 percent of hiring managers believe that high school graduates are less prepared to enter the work force compared with previous generations.

There are few high-quality studies on the impact of artificial intelligence on K-12 students and teachers, and the results of the studies that exist are mixed. Stanford’s A.I. Hub for Education recently published a review of over 800 academic papers and found that “A.I. tools may help students complete tasks more successfully in the moment, but those gains do not always persist when students are later asked to perform independently.”

Even as the gains are limited, the pitfalls are mounting. Researchers at the University of Pennsylvania have warned about the risks of “cognitive surrender” to artificial intelligence among the general population, which is when users of A.I. avoid “effortful thinking” and offload a significant part of their decision-making to large language models. What’s more, it is possible that whatever is being taught to schoolchildren today could be obsolete in six months.

When I asked leaders throughout the Seckinger cluster about how they were defining the success of their A.I. pilot, they were refreshingly honest about the rocky terrain beneath them.

Seckinger doesn’t have an ultimate goal it’s trying to meet, or metric it’s trying to measure, said Sallie Holloway, who is the director of artificial intelligence and computer science for Gwinnett County Public Schools and who helped develop the learning framework for the cluster. She started her career as a computer science teacher and instructional coach and stepped into her current role in 2021.

It’s more like a “system of metrics,” she said, that include normal public school student achievement like grades and test scores and whether students are building technological skills and feel positively about the curriculum. The experimental nature of what is being attempted in Gwinnett County reminded me of an oft-repeated start-up maxim: Seckinger is building the plane while flying it.

- Author: Jessica Grose, The New York Times

The post will continue...

Read more...