An Overview of Research on AI in Education

An Overview of Research on AI in Education

An Overview of Research on AI in Education

An Overview of Research on AI in Education

5

Min read

Aug 26, 2024

Aug 26, 2024

Share this

Share this

Share this

Share this

Share this

Remember: we always welcome your feedback.

Featured Articles: Introducing #PlatformPitches

#PlatformPitches is the name we’ve given to our new series of articles, offering suggestions to the Republican ticket. There are now three articles in the series:

Introducing #PlatformPitches: Our Suggestions for the Republican Ticket

#PlatformPitches: Time For A New COPS Program

#PlatformPitches: You Want Price Controls? I’ll Show You Price Controls

NEW: #PlatformPitches: More Paid Vacation Time

An Overview of Research on AI in Education

We appreciated this post by John Bailey of AEI summarizing the most recent research on the uses of AI in education. You should read the whole thing but here are the highlights:

AI as an Efficient Grader: One of the most promising applications of AI in education is its ability to grade short-answer questions. Recent studies have shown that advanced AI models like GPT-4 can grade almost as accurately as human teachers, but with significantly greater speed. This efficiency could potentially free up substantial time for educators to focus on other critical aspects of teaching. However, the research also highlights the need for further investigation into the types of questions AI can effectively evaluate and the datasets used to train these models.

The Double-Edged Sword of AI Tutoring: AI-powered tutoring tools have demonstrated impressive results in improving student performance during practice sessions. However, the research reveals a concerning trend: students who relied heavily on AI assistance showed decreased performance when the tool was unavailable during exams. This suggests that AI tutors aren’t ready for large scale rollout yet.

Predicting Social Science Experiments with AI. We’ll just quote directly, here: “Finally, a fascinating study examined GPT-4’s ability to predict the results of social science experiments. Analyzing 476 treatment effects from 70 pre-registered survey experiments, researchers found that LLM-derived predictions strongly correlated with actual effects, even for unpublished studies. LLMs matched or surpassed human forecasters’ accuracy and showed consistency across demographic subgroups. In nine additional “mega-studies” testing various interventions, LLM predictions were less accurate but still comparable to expert forecasts, especially for text-based survey experiments.”

More at the link.

Policy News You Need This Morning

#FreeSpeech #BigTech – Mark Zuckerberg’s bombshell letter to the House Judiciary Committee is getting a lot of tongues wagging, and rightly so. In the letter, Zuckerberg admits being “pressured” by the Biden-Harris Administration to censor Americans and admits to caving to that pressure. More.

#FreeSpeech #Academia – Speaking of free speech, a threat to free speech and academic liberty is scholarly associations making political statements. The chilling effect this has on academic speech should be obvious. In a very welcome report, AEI’s excellent education expert Frederick M. Hess looks at this situation we shouldn’t take for granted and calls for a straightforward solution. Choice quote: “Of the 99 academic associations examined, 81 percent have issued at least one official position on race or affirmative action, the Russian invasion of Ukraine, the Israel-Hamas conflict, immigration, or climate change. These statements almost uniformly reflect progressive orthodoxy. We estimate that public colleges and universities spend nearly $200 million a year subsidizing faculty dues and paying conference registration fees to politicized associations. While faculty certainly have a right to participate in such organizations, they have no right to do so with public funds.” Seems like common sense to us…

#FreeSpeech #Academia – Speaking of free speech and scholarly associations, from WSJ: “The American Association of University Professors on Aug. 9 threw gasoline on the still-smoldering embers of this past spring’s campus encampments. The organization withdrew its nearly two-decades-old opposition to academic boycotts, in which faculty, academic institutions and scholarly organizations refuse to work with certain institutions and scholars.” If they get any public money, it should be withdrawn immediately. At a minimum.

#PublicHealth – AEI’s excellent public health expert Scott Gottlieb is in WaPo talking about a subject we’d all rather stop thinking about, but which we shouldn’t: dangerous things occasionally escape from biolabs. We can’t resist quoting the juiciest part: “In 1977, a novel strain of H1N1 influenza, which had largely vanished from human circulation, suddenly resurfaced and sparked a pandemic, believed to have been the result of human experimentation with the strain, in either Russia or China. The following year, a medical photographer in England contracted smallpox and died after the virus leaked out from an experiment. In 1979, an anthrax outbreak hit Yekaterinburg, Russia. The spores, spreading from a military microbiology facility, went on to claim at least 68 lives. In 2019, brucellosis broke out in China’s Gansu province after a pharmaceutical plant used expired disinfectants, allowing a large plume of the bacteria to slip through exhaust vents and waft into nearby neighborhoods. Staggeringly, approximately four months passed before the public was informed. By then, at least 10,000 people had been exposed, with hundreds falling ill.” That sounds bad. Policy should address it. In any case, we’ll be over here with our “Wuhan Institute of Virology” t-shirt.

#Reg – You have read us cover the Biden Administration’s nearly-unprecedented regulatory bulimia. At City Journal, Casey B. Mulligan makes a very good observation: those new regs are not just economically harmful as a whole, they have also been “highly regressive, with low-income households bearing a large cost as a share of their income.”

#Immigration – It’s a tough reelection campaign for Justin Trudeau. So tough that the most woke, pro-immigration leader in the West admitted that immigration may have downsides. American Compass’s Oren Cass rewards him with highly-merited snark: “Wait, so Canada’s experience with a flood of low-wage, temporary foreign workers wasn’t a booming labor market for Canadian workers and youth?! But, but, economists have spent so much time developing convoluted models for why this works out great. You mean they didn’t pan out?” Fair question.

#Immigration – Speaking of… No comment: “If she’s elected president, Kamala Harris pledges to spend hundreds of millions of dollars on the wall along the southern border — a project she once opposed and called ‘un-American’ during the Trump administration.” (Via)

#AI – Interesting, from RAND: “By some estimates, more than 80 percent of AI projects fail — twice the rate of failure for information technology projects that do not involve AI. Thus, understanding how to translate AI’s enormous potential into concrete results remains an urgent challenge.” More.

#FinReg – Interesting from a new NBER paper: “Stock prices of U.S. payment firms decrease, while stock prices of European payment firms increase in response to positive announcements on the digital euro. Bank stocks do not react.” The authors conclude that geopolitics cannot be separated from the issue of central bank digital currencies (CBDCs).

#FinReg #Crypto – Nasdaq is seeking SEC approval to launch options on a Bitcoin index, Cointelegraph reports. More integration of crypto into the regulated financial economy.

Chart of the Day

Not a single country in the entire world has a higher fertility rate than in 1950. (Via)

Meme of the Day

PolicySphere

Newsletter

By clicking Subscribe, you agree to share your email address with PolicySphere to receive the Morning Briefing. Full terms

By clicking Subscribe, you agree to share your email address with PolicySphere to receive the Morning Briefing. Full terms

PolicySphere

Newsletter

By clicking Subscribe, you agree to share your email address with PolicySphere to receive the Morning Briefing. Full terms

By clicking Subscribe, you agree to share your email address with PolicySphere to receive the Morning Briefing. Full terms