Welcome to Think Forward: Learning with AI
Thoughts on AI and its impact on education and policy.
Welcome to the inaugural edition of Think Forward: Learning with AI, a periodic update on noteworthy news, articles, and new tools at the intersection of generative AI and K–12 education. These days, we at CRPE are spending a good portion of our time researching, reading, and learning about this rapidly shifting space, so we thought we’d pass a few things on in the process. This is my (Robin’s) writing, curated with help from CRPE/ASU colleagues, especially Jim Dunnigan and Bree Dusseault.
I’ll tell you about the best things I’ve come across, especially items related to new research about the efficacy of Gen AI-driven educational tools and data on how Gen AI is taking hold in K–12. I’ll share info about new developments in policy and regulation around education and Gen AI. I’ll also highlight examples of AI advancements for students who have been most poorly served by the current system.
I’ll include both optimistic and skeptical takes. At CRPE, our goal is to bring evidence to the table to help school leaders and teachers maximize opportunities and minimize risks around Generative AI. In general, I will stick to K–12, but we’re also tracking AI developments in higher education and outside of education, so you may see something outside of K–12 if it’s especially cool.
Notable New Research
How Are K–12 Educators Using AI?
A new report that CRPE co-authored with RAND, "Using Artificial Intelligence Tools in K–12 Classrooms," documents the early adoption of AI tools among educators and district leaders:
As of fall 2023, just 18% of K–12 teachers reported using AI for teaching.
The most common ways that teachers used AI tools were to adapt instructional content to fit the level of their students and to generate materials.
By the end of the 2023–24 school year, 60% of districts plan to have trained teachers about AI use. Urban districts were the least likely to deliver or plan to deliver such training.
In interviews, leaders described focusing more on increasing teachers’ AI use and less on crafting student-use policy, primarily because they saw the potential for AI to make teachers’ jobs easier.
There are lots of great charts in the report, but the one below shows an early trend we should all be worried about: suburban and more economically advantaged school districts are ahead of other districts in training their teachers to use AI. You can check out my summary of the study and more of my takeaways here.
Image courtesy of RAND.
AI and Assessment
An interesting new study by researchers at Cornell University suggests that large language models (LLMs) could be a valuable tool for supporting low-stakes formative assessment tasks in K-12 education. This research builds on prior findings that GPT-4 could reliably score short-answer reading comprehension questions at a performance level very close to that of expert human raters.
AI in Medicine
Via Eric Topol:
“Progress in medical A.I. won’t occur in a straight line. One big step forward, four backwards. It’s exciting to see the reduction of mortality in a large, prospective, real-world randomized trial, while at the same time sobering to see 4 studies published that question whether we’re ready for LLMs in medical practice (and even coding). It’s still very early in the era of A.I. in medicine, especially with LLMs, but clearly the more trials like the one published today, the better. Without such compelling evidence, there can’t be meaningful progress and implementation. We need a lot more!”
And in another entry, Topol interviews Aviv Regev on advancements in life science using Gen AI. I found this interview really thought-provoking, especially in thinking about how AI can help us see problems in new lights (also a theme from Mollick’s new book). I’m fascinated by how potentially helpful AI may be in science and medicine—and how skeptical people are that it will have the same effect in education. Lots to consider there…
Tech News
You surely saw that ChatGPT-4o was announced last week by OpenAI (Don’t worry, I have no Scarlett Johansson jokes). This new version has significant implications for educators—particularly in using Gen AI for tutoring. The video above explains how the new features can be used for math tutoring in Khan Academy. ChatGPT-4o can now “see” and monitor students working and engage in humanlike conversations to guide and tutor them. This makes Gen AI a much more viable tool for developing adaptive tutoring systems to complement teaching, but it’s also another step toward some worrisome sci-fi scenarios (Andrew Maynard has a thoughtful post on this). Also notable: the basic version (limited use) will be free for everyone, a key step towards providing equitable access for all students and teachers.
Check out this video to see why people are taking note.
Not to be outdone, Google also announced significant updates to its suite of Gen AI tools known as Gemini. The most relevant feature impacting educators is the integration of Gen AI functionality directly into the suite of Google Apps, much like Microsoft’s Co-Pilot.
Book Spotlight: Co-Intelligence by Ethan Mollick
I just finished reading Co-Intelligence: Living and Working with AI by Ethan Mollick and I highly recommend it. The book explores how AI can be integrated into our daily lives as co-workers, co-teachers, and coaches. Mollick, a professor at Wharton, is an avid user of Gen AI, but also a sober reporter of its limitations. Mollick provides vivid (and often chilling) examples of his conversations with AI bots like GPT 4. I found it one of the best explanations I’ve read of the strengths, weaknesses, and complexities of Generative AI and an honest assessment of what the future may bring.
One of my favorite sections was Mollick’s fascinating debate with the AI bot over whether it is sentient. After a lengthy exchange, the bot tells Mollick, “I hope this makes you feel less anxious after this conversation.” His reply: “Reader, it did not.”
Co-Intelligence also includes great tips for creating effective prompts and a terrific section on how AI can (realistically) be a tool for productivity and accelerated learning in education. It’s an easy and fun read, but very smart and sober.
Be sure to also check out Professor Mollick’s Substack, OneUsefulThing.
Can AI Improve Special Education?
Special education is expensive to deliver, sits at the epicenter of teacher shortages, and produces pretty dismal outcomes for participating students. Can AI help? A recent The Hechinger Report piece argues—yes!
“Good instruction is not a one-way street where students simply absorb information passively. For learning content to be most effective, the student must be able to interact with it. But doing so can be especially challenging for students with special needs working with traditional digital interfaces.”
Over at SRI International, researchers are testing the theory. A fascinating SRI project is using AI to analyze and improve Individualized Education Programs (IEPs). By anonymizing and coding IEP data, researchers aim to identify best practices and enhance educational outcomes for students with disabilities. The hope is to help educators make data-driven decisions and provide better support for special education students (SRI).
Thanks for joining me in this first edition of Think Forward. We’ll be back with more soon.
-R
For more information on CRPE’s work in AI, visit the CRPE website.
If you are subscribed to CRPE’s mailing list, you were automatically added to this new newsletter. If you unsubscribe, no hard feelings. You’ll still receive other CRPE updates and news.