March 2024: The research on AI for education is in
eSpark Learning
Engaging instruction and practice for reading, writing, and math in grades PK-8 ??
In this month's edition of EdTech Evolved , we're sharing insights from educators and researchers all across the country. A new survey paints an impressive picture of teachers' growing AI adoption, while researchers at 微软 published the results of one of the first large experiments on the impact of AI for math instruction. Plus, get advice from a leading provider of AI trainings on how you can AI-proof instruction, and explore the first-ever AI Impact Assessment Template for K-12 Schools .
Survey Report: The State of AI for Teachers
Today, nearly every educational conference has keynote speakers discussing artificial intelligence. New products and tools are being released every month. But how are teachers really feeling about AI, and how much progress have they made integrating it into their teaching practice? We sent a survey to thousands of U.S. teachers, and the results shed light on the remarkable journey they're taking.
In just one year, we’ve witnessed a nearly 2x leap in teacher comfort with using AI in the classroom, from an average rating of 2.9 (out of 10) to 5.6. But this growth is not a product of structured training programs — 66% of teachers report they haven’t received any formal instruction on leveraging AI for education. Only 8% reported trying AI at their administration’s request, underscoring the current grassroots nature of this adoption.
That said, the path to fully integrating AI into teaching practices is still unfolding. Teachers today rate their experience with AI as 3.3 out of 5, indicating a cautious optimism. Most are experimenting with AI occasionally, with 73% only using it a couple times a year or monthly. This suggests we’re only beginning to scratch the surface of what true classroom integration could look like.
Research Roundup: Promising Evidence on AI for Math Instruction
With more and more teachers bringing AI into their classrooms, the question remains: is this actually good for learning?
As with any new breakthrough, the research has taken some time to catch up. But now, researchers from 微软 have finally published the results of one of the first large, pre-registered experiments exploring the efficacy of AI for math instruction.
The study looked at how participants performed on problems designed to mirror SAT math questions. Each participant was randomly assigned to one of six possible conditions to identify whether trying a practice problem, seeing the correct answer, and getting "stock" or "customized" explanations from GPT-4 helped them successfully answer two final test problems.
The results give us a strong indication that LLMs like GPT-4 can add meaningful value by playing the role of real-time tutor: GPT-4 was able to equip subjects with the strategies they needed to improve their results in a span of mere minutes.
How to Redefine Learning in an AI-Powered World
领英推荐
William Grube , founder of Gruvy Education , works with schools nationwide to train educators on how they can integrate AI tools into their teaching practice. He's seen firsthand how the AI era is redefining learning and is sharing tips on how you can get ahead of the curve.
The best place to start for most of the teachers I’ve worked with is by understanding the limitations of AI. Begin every lesson by asking yourself “Could a student use ChatGPT to complete this assignment in its entirety?” If so, it’s safe to assume that at least some of your students are doing just that, leading them down an endless loop wherein the only skill they’re developing is how to cut corners more effectively.
As William sees it, educators can terminate the arms race between cheating and anti-cheating technology and make it impossible to cheat by changing the way material is presented. This approach not only increases trust and integrity in the classroom, it also helps students learn how to use AI productively rather than covertly.
The AI Impact Assessment Template for K-12 Schools
From the EU’s Artificial Intelligence Act , to the White House Blueprint for an AI Bill of Rights , to the growing volume of state legislation being enacted, the recommendation for organizations to conduct what’s known as an “impact assessment” before implementing certain types of AI tools is everywhere. And yet, we could find no education-specific examples of what such an impact assessment might look like.
Thus, the EdTech Evolved AI Impact Assessment Template for K-12 schools was born.
When should you use the AI Impact Assessment Template?
Generally speaking, the states that require impact assessments mandate them for tools that automate decision-making. This includes basically any adoption of AI that could have a disparate impact on individuals or groups based on algorithmic decisions. In K-12, this might include things like automated special education or behavior referrals. It could also include college and career recommendations, school lotteries, attendance zoning, and automated class assignment.
That's a wrap on March's EdTech Evolved! Subscribe at the top of this page to make sure you catch next month's edition.
Check out more EdTech Evolved articles here , and comment to tell us which parts of this newsletter you enjoyed so we can keep bringing you the best updates on AI in education.