AI for Literacy

The intervention consists of guided activities in which early grade literacy teachers engage with structured scenarios that simulate the use of artificial intelligence (AI)–supported instructional responses. Grounded in the Technology Acceptance Model (TAM) and AI literacy frameworks, the activities invite teachers to analyze, interpret, and evaluate AI-generated outputs related to literacy practices, including instructional materials and reading analyses. Depending on random assignment, participants complete the activities either before or after responding to a questionnaire.

The intervention consists of a structured sequence of facilitator-led activities designed to simulate interaction with an artificial intelligence (AI)–supported tool for literacy-related instructional tasks. The intervention does not involve the use of a live AI system. Instead, it relies on standardized prompts, pre-defined AI-like outputs, and guided discussion to reproduce key features of AI-supported decision-making. The activities are grounded in the Technology Acceptance Model (TAM) and AI literacy frameworks, explicitly targeting constructs related to (i) understanding how AI systems generate outputs, (ii) evaluating the quality, limitations, and biases of AI-generated content, and (iii) applying human judgment to mediate and adapt AI-supported instructional suggestions.
Across activities, teachers engage with simulated scenarios that include: (a) classifying which tasks are more appropriately handled by teachers, by AI, or by a combination of both; (b) critically evaluating AI-generated instructional texts and materials for phonological accuracy, pedagogical adequacy, and contextual relevance; (c) analyzing simulated AI-generated literacy assessments or reading analyses under different instructional themes (e.g. fluency, comprehension, vocabulary); and (d) identifying limits, errors, simplifications, or biases in AI-generated recommendations. Participants are prompted to discuss what aspects of the outputs appear useful, what requires revision, and what remains dependent on professional judgment and contextual knowledge. Activities are delivered in a group setting outside regular classroom instruction and last approximately 90 minutes per session.

2026-04-09

2026-05-31

Perceived ease of use of AI for teaching-related tasks
Trust in AI as support for pedagogical work
Behavioral intention to use AI with guidance
Perceived usefulness of AI for key teaching-related tasks (planning, materials, assessment, instructional strategies, administrative tasks)

Primary outcomes are measured through a structured questionnaire administered to early grade literacy teachers. All primary outcomes use 4-point Likert-type response options (from “Disagree” to “Agree” or equivalent).

Self-efficacy to learn new digital tools for literacy support
Perceived school support for digital technology use
Main perceived barrier to using AI in teaching-related work
Self-reported practices for reviewing/adapting AI-generated content
Self-reported data privacy and caution when using AI with student data
Prompting practices (specificity and iterative refinement when making AI requests)

Secondary outcomes are measured through additional questionnaire items capturing enabling conditions and self-reported practices.

The study uses a randomized controlled design in which participating literacy teachers are randomly assigned to receive the intervention either before or after completing a questionnaire. This design allows for comparisons between teachers who have been exposed to the simulated AI-based activities at the time of assessment and those who have not yet participated. All participants ultimately receive the intervention.

The study employs a randomized controlled design with individual-level random assignment among early grade literacy teachers. Participants are randomly assigned to one of two experimental arms: (i) an early-intervention group, which completes the intervention prior to responding to the questionnaire, and (ii) a delayed-intervention group, which completes the same questionnaire before exposure to the intervention. This before–after design allows for estimation of short-term differences in measured outcomes associated with exposure to the intervention, comparing teachers who have engaged with the simulated AI-based activities at the time of survey completion to those who have not yet participated. All participants ultimately receive identical intervention content, ensuring equity and minimizing ethical concerns related to withholding treatment. Randomization is conducted using a computer-generated assignment procedure by the research team. The unit of randomization is the individual teacher, and no clustering is applied at the school or municipal level.
Outcomes are measured immediately following survey completion and reflect participants’ perceptions, intentions, and self-reported practices related to AI use. Given the short time horizon between intervention and measurement, the design is explicitly focused on capturing immediate shifts in perceptions, confidence, perceived usefulness, and reported practices, rather than long-term behavioral change or student learning outcomes.

Randomization done in office by a computer

Individual (literacy teachers)

No

No

200 primary teachers

100 teachers with intervention before questionnaire, 100 teachers with intervention after questionnaire (control at time of survey)