What Instructional Leaders Should Ask Before Adopting AI for Literacy
AI tools can scale coherent literacy instruction — or further fragment it. Here’s how to tell the difference.
AI tools for classrooms are arriving faster than anyone can evaluate them. Districts that spent years strengthening literacy instruction — adopting structured programs, retraining teachers, and improving curriculum alignment — are now layering AI tools on top of that foundation faster than anyone has figured out how to evaluate them.
Recently, the LitLab team joined a conversation hosted by Student Achievement Partners on exactly this issue. (Full webinar here). In that discussion, instructional leaders and educators were asked to describe their experiences with AI in the classroom. A wide range of answers were given: "overwhelming", "disjointed", "exciting", "unregulated", "beta mode", "conflicting emotions". That range is something we hear constantly from teachers and administrators as schools are flooded with new AI tools with varying quality and curriculum alignment.
The coherence problem already existed before AI arrived
For anyone working in foundational literacy, the coherence problem isn't new. It's the same problem that arose when districts adopted strong tier-one phonics programs while simultaneously using leveled readers for independent practice. The instruction was solid, but the practice materials didn't match it. And the students who most needed practice applying taught phonics patterns were often practicing text that introduced untaught ones instead.
Research is increasingly clear that alignment is not a minor implementation detail — it meaningfully affects reading outcomes. In a randomized controlled trial in Knox County, Tennessee, students receiving tutoring aligned to their Tier 1 classroom instruction outperformed peers receiving tutoring from a different program by 0.12 standard deviations — roughly 1.3 additional months of learning — despite both groups receiving high-quality, high-dosage support. For students in the bottom 10% of performers, the effects were even larger.
The mechanism behind that gap isn't mysterious. Saha et al. (2021) found that the level of decoding difficulty in a text is a unique and significant predictor of both reading miscues and passage fluency — meaning when students encounter phonics patterns they haven't yet been taught, accuracy and fluency suffer measurably, regardless of how strong the instruction was. Practice in misaligned materials doesn't just fail to help. It actively works against what the lesson taught.
More reading practice wasn't the variable in Knox County. More aligned reading practice was. AI doesn't change this math. It scales it in either direction.
AI as amplification vs. fragmentation
When AI is used as a curriculum amplifier, it extends the reach of the instruction a school has already invested in. It fills the practice gap with materials that are genuinely tied to a specific scope and sequence. It gives teachers data that traces back to what was taught that day, not a generic readability level. And it handles the tedious parts — materials creation, alignment checking, progress tracking — so teachers can focus their time where it matters most: working directly with students.
When AI is used without a curricular anchor, it can fragment instruction. A general-purpose decodable generator, for example, can produce beginner-friendly text, but it has no concept of a phonics scope and sequence. It doesn't know which sound-spelling correspondences your students have learned, which patterns are still ahead, or what lesson happened this morning. The result looks like a decodable, but it isn't. A text that sits at 10% decodability relative to a student's actual instructional level isn't decodable at all. It's a leveled reader in different packaging, doing exactly what the science of reading research tells us not to do.
A heuristic worth keeping
One useful frame for evaluating any AI tool: is it taking students deeper into your core instructional materials, or is it pulling them away from those materials?
It's not a rubric, but it's a heuristic that an instructional leader can apply quickly — in a walkthrough, a procurement conversation, or a PD session — and it cuts through a lot of noise.
This is the distinction that often gets lost in the rush to adopt more AI in reading instruction. For example, the question is not whether a tool can produce reading materials. The question is whether those materials are instructionally coherent with what students are being taught in the classroom.
What this means for instructional leaders
The schools that will navigate this moment best are the ones approaching AI adoption the same way they approached curriculum adoption: with careful questions about evidence, instructional coherence, and fit — not just excitement about new features.
That means asking vendors how their tools align with your Tier 1 program. It means asking for decodability scores against your actual scope and sequence — not a generic readability level. And it means being precise about the problem the tool is solving, and whether that problem is genuinely constraining student reading growth.
The AI tools that belong in literacy classrooms are the ones built to support coherent, Science of Reading–aligned instruction. For independent early reading practice, that means being grounded in the understanding that a phonics lesson is not complete until students have practiced the taught patterns in connected text that is deeply aligned with their teacher’s scope and sequence.
References
- Saha, N., Grieve, R., & Lovelock, I. (2021). Decoding difficulty as a predictor of oral reading fluency and miscues. Reading and Writing.
- Tennessee SCORE + Accelerate (2024–25). Knox County tutoring alignment RCT. (Confirm full citation with Varun.)
- Griffith, D. & Fitzpatrick, B. (April 2026). From the Teacher's Desk: A Science of Reading Progress Report. Thomas B. Fordham Institute. https://fordhaminstitute.org/national/research/teachers-desk-science-reading-progress-report
--
LitLab serves 5000+ school sites and 400,000+ students across K–5. It is ESSA Tier IV certified and aligns to UFLI Foundations, CKLA, Fundations, Reading Horizons, Superkids, Benchmark, and other major phonics programs. 👉 Learn More to see how it fits with your current program.