Prospective founding contributors: Graham Horn, Stephen Jennings, Jacqueline Jacob, Ian Mauer, and Paul Consalvi
Open invitation to: language teachers, writing instructors, students, AI-literacy researchers, educational technologists, assessment specialists, and institutional policy contributors
Generative AI can expand ideas, explain concepts, restructure prose, simulate counterarguments, and offer rapid feedback. The same tools can also displace productive struggle, obscure authorship, create unearned confidence, and weaken a learner’s ability to judge without assistance. The important question is no longer whether AI should be used. It is when, why, and under what conditions AI strengthens human reasoning rather than replacing it.
We invite collaborators to form a cross-institutional action-research group investigating AI as a scaffold, thinking partner, feedback mechanism, and potential source of critical-thinking dependence.
To develop evidence-informed pedagogies in which AI increases learners’ capability, agency, and accountability rather than merely improving the appearance of their products.
Define observable differences between AI-supported thinking and AI-displaced thinking.
Identify when AI is most useful across brainstorming, research, drafting, feedback, revision, and reflection.
Develop structured protocols requiring learners to question, compare, verify, select, adapt, and justify AI output.
Examine how AI use affects authorship, confidence, metacognition, disciplinary learning, and independent performance.
Design feasible classroom evidence that focuses on process as well as final product.
Develop guidance that can adapt to rapidly changing tools without depending on one platform.
A multi-institutional evidence base for human-agency-centered AI pedagogy.
A practical continuum distinguishing substitution, assistance, scaffolding, partnership, and dependence.
Teaching and assessment designs that preserve human judgment while using AI’s genuine advantages.
Student-facing AI learning contracts that make purpose, permitted support, disclosure, verification, and accountability explicit.
A shared action-research dataset and a series of publishable classroom studies.
Human agency: AI can support decisions; it does not inherit responsibility for them.
Process transparency: Learners should be able to explain where AI entered the work and what they did with its output.
Capability growth: Assistance should help learners become able to do more, not permanently require more assistance.
Verification: Fluency and confidence are not substitutes for accuracy, relevance, or evidence.
Proportionality: The depth of disclosure and control should match the stakes and purpose of the task.
Learner voice: Students should contribute evidence about what helps, what displaces effort, and what feels fair.
What observable evidence suggests that AI is scaffolding rather than substituting for thinking?
Which forms of productive struggle should teachers deliberately preserve?
Does structured AI brainstorming improve the diversity and quality of options learners later evaluate?
How does repeated AI assistance affect independent performance when assistance is removed?
How should teachers evaluate a strong product when the learner cannot defend its reasoning or choices?
What forms of AI disclosure support learning without turning the classroom into surveillance?
An AI Support–Agency Continuum with classroom examples.
A reusable Ask AI → Question → Verify → Decide → Reflect protocol.
A student AI learning contract and process-log template.
A common action-research design with adaptable pre-, during-, and post-task measures.
A repository of tasks designed to compare unstructured AI use, structured AI partnership, and independent performance.
Weeks 1–3: Agree on working definitions of agency, scaffolding, substitution, and dependence.
Weeks 4–6: Select one common task and design locally appropriate comparison conditions.
Weeks 7–10: Run small classroom pilots and collect process evidence, products, defenses, and learner reflections.
Weeks 11–13: Compare patterns, identify methodological limitations, and refine the shared protocol.
Classroom access, student co-researchers, action-research expertise, process-writing pedagogy, AI literacy, qualitative analysis, assessment design, ethics and policy knowledge, and longitudinal research capacity.
Create human-agency, research-ethics, classroom-practice, and methods stewards. Treat AI policies and tools as changing conditions. Maintain a decision log so that the group can distinguish evidence-based revisions from reactions to new technology or public anxiety.
Join if you want to investigate a practical and consequential question: After working with AI, what can the learner understand, judge, explain, and do more capably?