Out of the ashes: What comes next for education—and how can AI help?
Plus more on an integrated AI school strategy, some terrific commentaries and interviews, and an upcoming webinar (you're invited!)
It’s been a busy few weeks in the education world, what with the goings on over at the U.S. Department of Education and the Institute of Education Sciences (IES)/National Center for Education Statistics (NCES), the agencies that support education data and research. I’ve been spending a lot of my time thinking about what kind of phoenix could rise from the ashes if the current functions of these federal education agencies are rebuilt and rethought (hopefully). One such “phoenix rising” could be a really smart large-scale investment in AI education research and development. IMHO, a re-conceived IES should prioritize questions like:
How can AI leverage evidence-based practices in education?
Which AI tools demonstrate efficacy, and with which types of students (e.g., self-motivated learners, etc.)?
What are user (teacher, student, parent, principal) experiences with and feedback on AI tools?
How can AI improve education for students with unique needs (like students with disabilities or English learners, for example)?
What are innovative ways to use AI to improve efficiency and data analysis and create cost savings in school district operations?
What effect does AI policy have on school- and classroom-level innovation? Which policies are most effective in mitigating risks, and which ones are ineffective and inhibit innovation?
How could AI help transform the teaching profession so educator time is put to the highest and best use in service of student learning?
How could AI help parents be more informed consumers of school choice and better advocates for their children, and how can AI help them get a more accurate understanding of their children’s academic proficiency?
The new IES could drive a lot of important innovation and improvement in AI and education by answering questions like these and becoming an aggressive arbiter of “what works” as we enter this brave new world of generative AI in education. On the other hand, a forever diminished IES could hobble our progress toward the future.
What shape would you like to see a revamped federal role in AI and education take? What AI-related questions would you like to see answered? Drop me a comment and let me know.
A quick follow-up from my last post on integrated AI school strategy
My last post inspired a lot of great feedback and was one of my most-read entries. I’m glad many of you share my interest in an integrated AI school strategy.
Folks sent good thoughts for sharpening specific recommendations in my R&D agenda and offered interest in bringing together funders, educators, edtech leaders, and policy people to design new AI-powered school models. One reader suggested that I should do more to help people—especially those in the edtech sector —understand why integration and coherence are so important in schools. I’ll be writing more about that, but here are some initial thoughts…
Teachers and students operate within a set of constraints that can inhibit goals like personalized learning for all, even with the best instructional tools. The current, outdated delivery model is hardwired for sameness and is built around rigid structures that do not always serve students. Examples of such rigidities include scheduling, seat-time requirements, pacing guides, “grade level” assessments and accountability requirements, and data systems that hinder tools designed to let students learn at their own pace, to provide intensive tutoring, to backfill missing knowledge, or to accelerate students into higher-level learning.
Teacher professional training and development is often under-resourced, misaligned with organizational goals, and poorly delivered, leading to teacher mistrust of new curricular approaches and tech tools.
Education systems frequently suffer from siloed programs (e.g., special education, bilingual education, career-technical programs) that do not work together to serve students, as well as a host of softer school- and district-wide detrimental norms and structures, such as tolerance for low expectations and ineffective approaches to student behavior and culture.
All of these dynamics can work alone or in concert to undermine coherence and integration—and ambitious visions of deeply personalized and customized learning driven by AI, “anywhere-anytime” learning, mastery-based learning, more relevant career pathways for students, or even just the adoption of AI-powered, evidence-based tutoring tools.
Here are some more ways that this lack of integration undermines educators’ ability to do the things we know work for young people:
Traditional district structures, with layers of compliance-driven regulations, often hinder coherence by making it difficult for schools to integrate innovative practices across departments and levels of governance. Inconsistent state policies can make things worse.
All schools, public or private, operate in political environments. This can play out in rigid union contracts, counterproductive school board policies, and many other ways that undermine whole-school improvement efforts and uptake of effective practices.
Systems might use multiple AI-driven platforms that don’t communicate or work together, resulting in inconsistent student experiences. A district might use an AI-powered reading assistant in English classes, an AI math tutor in another, and a separate platform for grading—all with different data standards and levels of teacher training. Teachers may distrust or resist adopting new AI-related programs. Getting beyond the “first adopter” problem in schools is an evergreen challenge and cannot be overcome without integration strategies.
My research center was founded on the concept (based on decades of effective school research) that breakthrough student outcomes depend on coherent schools and school systems that integrate various components—curriculum, instruction, professional development, funding, staffing, accountability, and governance—into a unified strategy that supports student success. I believe AI can help us achieve transformative outcomes, but not without integration. No tech company would operate without coherence; why should we expect schools to do so?
At CRPE, we’re actively studying AI “early adopter” school districts to document what they are learning and what barriers they face. My colleagues will share early findings from CRPE’s new study of school districts and charter management organizations leading the way in intentional AI implementation at an upcoming webinar on March 13. We welcome you to join the discussion.
In Other News
This new commentary by Robin Hood Foundation’s Amber Oliver is full of terrific examples of how Robin Hood is investing in NYC innovations to address and overcome inequalities:
For decades, students of color and those from low-income communities have faced persistent achievement gaps in our education system. Despite countless reform efforts, these students continue to encounter barriers to accessing high-quality, personalized instruction that builds critical thinking and problem-solving skills. Now, the emergence of generative AI represents an unprecedented opportunity to transform this inequitable system into one that truly serves all students.
This piece is the second in our ongoing AI Learning Series, where CRPE invites thought leaders, educators, researchers, and innovators to explore the transformative potential—and complex challenges—of AI in education. Stay tuned for more.
New AI + Math Research Opportunity
This upcoming research partnership RFP from Digital Promise and The Gates Foundation is a terrific opportunity for researchers to dig into math data from leading AI education tool providers like Curriculum Associates, Khan Academy, UF Lastinger Center, and Math Nation (and, later this year, OpenStax).
The program especially encourages applications from graduate students and early career professionals, educators with research roles, such as those working in district research offices, and researchers located at minority-serving institutions or Title I school districts.
Some Needed Optimism
I really enjoyed this interview with Reid Hoffman (founder of LinkedIn) about his new book, Superagency. In contrast to Yuval Noah Harari, Hoffman is an optimist when it comes to AI, but I appreciated his thoughtful and sincere responses to the probing questions from one of my favorite podcast hosts, Russ Roberts of EconTalk. I’ll be reading the book soon and will report back with more thoughts. As always, I believe we must understand risks and design toward the opportunities. Plus, a little optimism is welcome right now, both in AI and in the world.
Kristen DiCerbo, Chief Academic Officer at Khan Academy, always has smart, instructionally grounded things to say, so I was eager to read her recent interview with Greg Toppo in the 74. I was especially interested in what she had to say about new developments in assessment, including conversation-based assessments, and how AI might be able to improve testing practices. Conversational testing seems to be getting a lot of attention in higher education as well right now. I’ll be interested to know where this goes. I share Kristen’s view that there is great potential for AI to better inform parents about their child’s performance.
The last thing I think is interesting is helping teachers and parents make sense of assessment data and get recommendations. Can AI help with that? Instead of getting this printout that says, “Your student got a 580 on this,” and you’re like, “What does that even mean? What should I do?” If you could have a conversation about that [with AI], that might be an interesting piece. We’ve been exploring that in something we have called Class Snapshots and recommendations that allow teachers to talk about their students’ Khan Academy performance. What else might they assign? How might they group students based on those kinds of things?
Final Thought
Did you notice that this year’s Super Bowl ads were all sort of…strange? You’re not alone—many of the big ad buys (at $8 million for 30 seconds) used AI to produce animations, storylines, or other ad content. Some of these AI-driven ads were also for AI products. Ironically, OpenAI’s first-ever Super Bowl ad animation was created entirely by humans.
Some argue that this signals a shift towards using AI as a storytelling tool and that consumers are more receptive to this than they have been in the past. Others argue that the AI-heavy ads felt overproduced or disturbingly uncanny valley-esque. Some viewers even reported that the ads made them less likely to consider using AI products. Either way, this was a good experiment in how AI-driven content sits with consumers. We’ll see what next year’s Bowl ads bring.
Robin! Thank you for sharing this with me (Jimmy McCue). One of the key questions you raise—how AI can improve education for students with unique needs, of which you've named one example (English learners)—is of particular interest for me right now. As we discussed recently, there is no "average" student, although the typical statistician reports with such focus, and as a result, most educators anchor to this misguided notion of the median.
I've spent the last month reflecting on learners similar to those that I worked with in South Seattle (immigrant and refugee), those in San Diego County (many of which are MLLs), and those from my first experience in education (learners in a racially segregated town in Mississippi). It has reignited a passion in me to act more intentionally in my professional efforts to create cross-community, student-driven learning experiences, where students from different regions collaborate on real-world challenges ... while leveraging AI as a tool, not a crutch.
Frankly, I'm interested in exploring how AI could be used by a number of different unique student groups as a powerful tool for bridging language gaps, providing real-time coaching, mentorship, and tutoring. Could we see more personalized instruction (data shows math in AI has yet to be beneficially engaged due to the nature of the subject as problem solving, not technically regurgitating), ensuring that these students in particular can meaningfully engage in interdisciplinary, project-based work?
I also appreciated your mention of AI improving efficiency in school district operations, based on my experience at the SEA level. I’ve been reflecting on how AI-powered data analysis could not only transform competency-based assessments (so that we can transcend the standardized tests that continue to misguide our measurement of holistic student learning), but more importantly, provide real-time insights into student progress while freeing up educators to focus on mentorship rather than administrative tasks.
Are you familiar with any AI-driven platforms that could also help personalize learning pathways, allowing students to build agency over their own educational journeys?
I would love to engage with readers on the following questions that I need help in answering:
- How can AI be leveraged to reimagine portfolio-based assessment models as an alternative to standardized testing? I am a board member of SEEQS, which uses a portfolio defense model at the end of a student's 8th grade year. I find it powerful - how can AI continue to bolster it's benefit?
- What might a teacher-AI partnership look like, where technology enhances rather than replaces the deep relational work educators provide? I ask because many of the teachers that I work with are still reluctant, hesitant, or unsure of how to engage with AI. How do we make it easier for them to adopt and own this partnership?
- How can AI facilitate mentorship networks, connecting students with industry leaders and broadening access to career pathways? How do we make the "hustle" of networking less daunting and more accessible to those seeking alternative pathways to academic and professional credentialing and success?
Lastly, I couldn’t help but smile at your mention of Russ Roberts and EconTalk—I used to be an avid listener, but over the past few years, I've fallen off. Thanks for the reminder to follow this podcast! A similar good listen: Tyler Cowen's "Conversations with Tyler."
Looking forward to your next post and the upcoming webinar—excited to see where this conversation goes!