Back to School with AI, Part 3: Competency-Based Education and AI
Seven ways AI can support personalized, student-centered assessment
The more I think about generative artificial intelligence, the more I think about competency-based education.
Competency-based education (CBE) is a system around which school is designed. The vocabulary around it is, let’s just say, developing. You may know it as mastery-based or proficiency-based or performance-based or personalized, but the premise is the same: we should detach student learning from problematic measures like seat time and grades and standardized test scores. Instead, learning should be measured by authentic demonstrations of durable, transferable skills that students need to succeed in and beyond school.
My belief in CBE springs from my belief in what school can and should be: a learning environment where every student is empowered to do cognitively complex work that matters to them. I’ve been working on this for years, starting in 2016 when I worked at Global Online Academy and launched a CBE approach in its Student Program. Since then, I’ve worked with dozens of schools and learning organizations on integrating CBE into the culture, structure, and pedagogy of their communities.
AI’s potential as a copilot for learning, a tireless teaching assistant and data analyst who can learn from and act on student and educator needs, makes me hopeful for the CBE movement. Its most aspirational goals—personalized learning environments that disrupt inequities in education systems—are also its biggest implementation challenges. AI makes the customized assessments, flexible pacing, and student-led learning required for CBE feel more possible.
For this post, I thought about my experience with schools implementing CBE, their successes, and their challenges. I’ve organized my ideas according to the seven elements of CBE outlined in the Aurora Institute’s authoritative overview (an excellent introduction if you need it). To test these ideas, I worked with Anthropic’s Claude 2: I like that it can read and analyze documents, and it’s been well-reviewed for its synthesizing ability.
As I’ve tried to do throughout this series, I focused on practical applications we can try right now with existing tools. It’s a menu, not a checklist. I’d love to hear what you try and what you would add!
1. Students are empowered daily to make important decisions about their learning experiences, how they will create and apply knowledge, and how they will demonstrate their learning.
Student agency (defined beautifully in this post by Jennifer Davis Poon) should be embedded in the design of competency-based learning experiences. Students should have the time, space, and support to articulate their goals then pursue and reflect on those goals in learning experiences they lead.
How can AI help? I used a common starting point for inquiry-based learning, the United Nations Sustainable Development Goals (SDG) and Targets, and uploaded them to Claude. I asked it to help me develop a research project. Here’s my initial prompt:
I live on Cape Cod, and Claude’s initial suggestions were strong (at least three of them have been major topics in local media recently). I chose a project on coastal health related to SDG 14 (“Life Below Water”) and Claude was excellent at helping me flesh out the project: it outlined a weekly plan, suggested questions a teacher might ask me about my project proposal, and then refined the plan to align more specifically to one of the ten targets aligned to SDG 14. Here’s the whole interaction.
Claude’s responses end with suggestions on how it might continue to help, a feature that seems essential to encouraging a student to push their thinking. Students, especially those who are new to this kind of inquiry or project-based approach, need scaffolding and feedback to turn ideas into learning pathways, and this is an area of strength for AI. I found myself actually getting excited about the project as I used Claude’s suggestions to bring it to life.
2. Assessment is a meaningful, positive, and empowering learning experience for students that yields timely, relevant, and actionable evidence.
One way to think about assessment in CBE is that school should not be a place where students complete activities and courses assigned to them in order to be evaluated; rather, school should be a place where students build a body of work. They use their work to curate evidence of learning, to advocate for themselves, and to express who they are and their visions for their futures. They are partners in assessment, not subjects of it.
In this spirit, portfolio-based assessment is common practice in CBE schools. I think portfolios are ideal for an AI world. First, they are curated, designed, and defended by students, an inherently “AI-resistant” form of assessment due to the prioritization of student voice and choice. Second, AI can facilitate portfolio-based assessment by supporting student curation and reflection: online platforms can tag and organize student work according to criteria like competencies or content areas, and chatbots can offer feedback on drafts, use reflective prompts to help students clarify their thinking, and review student work to help students locate themes and patterns over the course of weeks or months.
3. Students receive timely, differentiated support based on their individual learning needs.
Timely and personalized feedback from a trusted expert is the most effective form of feedback. Yet, this level of support is onerous, if not unsustainable, for most teachers. It might not be, however, with the support of AI.
The researcher Dylan Wiliam said, “The only thing that matters about feedback is what students do with it.” I don’t think AI is ready to completely replace teacher feedback, but it is ready to help students act on it. I posed as a student looking to Claude to help me process and act on feedback I received from a teacher:
How many times have we received feedback and not been sure what it means or what to do with it? How many times have we, for lots of different reasons, not followed up with the person who gave us that feedback to seek clarity? How many times have we as givers of feedback felt frustrated that we couldn’t give every person in our care the level of personalized feedback they needed and deserved? We can be using AI right now to process and act on feedback.
4. Students progress based on evidence of mastery, not seat time.
A core tenet of competency-based education is that “learning is the constant, time is the variable.” Personally, I don’t yet have the confidence that AI can define, assess, and credential “mastery” on its own. However, I see immediate applications for AI to introduce more flexibility in time, place, and pace of learning for students, especially in formative assessments. This plays to AI’s strengths:
It can read assessments and generate alternatives at various levels of difficulty, which supports both differentiation and reassessment for students who are ready to extend their learning or students who need multiple attempts to demonstrate mastery.
It can offer feedback and review feedback when students need it, not just when teachers are able to provide it (see #3 above).
It can review prior work and make suggestions about what to learn next (see #5 below).
All of these applications relate to the nascent world of AI tutoring. For a deeper dive into AI and personalized learning, I recommend Sal Khan’s TED talk introducing his academy’s new tutoring tool, Khanmigo.
5. Students learn actively using different pathways and varied pacing.
It’s common in CBE schools for students to collaborate with an advisor to use prior assessments and work to shape personalized learning pathways. Could AI help me as a student come up with a learning plan? Could it make me as an advisor more efficient and more effective? I uploaded the last five blog posts I wrote when I was at Global Online Academy to Claude and asked it to propose a learning pathway based on my prior work. Here’s the prompt I used.
In just a few exchanges, Claude gave me a plan with suggested tasks, deliverables, and competencies. Its assessment of my previous work was, to my eyes, accurate, and I genuinely liked its suggestions. If I were a student, I would be happy about how quickly I could create potential ideas to bring to my advisor. If I were an advisor, asking a student to do this pre-work with AI would make me more effective and more efficient in helping that student design an individualized learning plan. Here’s the whole interaction.
6. Strategies to ensure equity for all students are embedded in the culture, structure, and pedagogy of schools and education systems.
CBE is an equity movement. Schools like Boston Day and Evening Academy and Big Picture Learning pioneered CBE out of a recognition that traditional, one-size-fits-all modes of education are not designed for individual learners and can perpetuate societal inequities that affect people from marginalized communities. CBE’s focus on the individual student requires a school to be an environment that adapts to the student more than it asks the student to adapt to the environment.
AI has powerful potential in this area, but if we want to use AI for CBE, we should prioritize access and inclusion. Schools should ensure students have access not just to the accounts and devices they need to use AI, but also to instruction and support that help them think critically about how AI responses are informed by design choices, algorithmic bias, and the content of datasets. In addition, these tools have already been shown to negatively affect people along racial, linguistic, and neurodivergent lines. Learning from the various ways different people are affected by AI should be a first step in determining how we want to use it in school. I’ll re-post this proposed AI Bill of Rights as a primer.
7. Rigorous, common expectations for learning (knowledge, skills, and dispositions) are explicit, transparent, measurable, and transferable.
CBE uses skills-based outcomes and rubrics throughout the learning process so that students and teachers can communicate using a shared learning vocabulary and so that students gain confidence in setting personal goals aligned to outcomes. This process begins with the development of a Portrait of a Graduate, a clear and learner-friendly articulation of the durable, transferable skills and traits our schools commit to developing in all students.
If your school has a Portrait, the first question is how to implement it. As I’ve written about before, AI is good at synthesizing information and abstract ideas and producing clear and usable learning goals and rubrics. I uploaded two pdfs to Claude: a Portrait of a Graduate from a school district close to where I live as well as the Next Generation Science Standards (NGSS) for high school physical science. Here’s the prompt and an excerpt of the response.
AI’s ability to connect Portraits to practice and its ability to review and reorganize curriculum, assessments, and learning goals could be a powerful way to use the Portrait as a lens to review the structure, culture, and pedagogy of your academic program.
Whether or not you’re “all in” on CBE, I hope this post has made clear that its elements are both good for students and essential to any high-quality education. This is where I hope our conversations about AI take us over the coming months: What do students need from a high-quality education? How do we know? How can AI help?
I’d love to hear how AI is helping you think about or implement CBE. Let’s talk more in the comments.
Links!
A long read, but this 2022 article, “Assessment in the Age of Artificial Intelligence” deepened my understanding of the long-term implications for schools.
Time Magazine’s 100 Most Influential People in AI is a useful, readable way to get to know the field.
Prof. Maha Bali on how she is adjusting her assessments this semester to account for AI.
I really liked this guide for parents on how to talk with their children about Al. In case you missed it, my last post was about how educators can talk with their students about AI.
Thanks for reading! If you enjoyed this post, I hope you’ll share it with others. It’s free to subscribe to Learning on Purpose. If you have feedback or just want to connect, you can always reach me at eric@erichudson.co.
This was my gut reaction in the winter. AI and CBE as two sides of a productive movement forward. With CBE in place, we can be less concerned about AI as a means of cheating. Authentic assessment makes such uses intrinsically less desirable, or so my thinking went at the time. Did you use CBE in your classrooms?
As ever, Eric, a really valuable distillation of wonders/provocations overlaid on an important frontier for schools (CBE). Thank you!