A recent national survey found that English and language arts teachers are enthusiastic about using artificial intelligence for personalizing reading comprehension and tailoring lessons to individual students. Personalized reading content, identifying learning differences, and making reading content interactive were high on their list of wants.
This enthusiasm is not misplaced. In classrooms across the country, large language models in education are showing the potential to dramatically change how students learn and practice literacy skills. These powerful AI systems, capable of generating high-quality text and engaging in life-like conversations, offer new ways to personalize instruction, foster student engagement, and support personalizing reading comprehension at scale. But with these opportunities come challenges. Most notably, educators must be thoughtful about how they implement these tools to ensure they promote equity, transparency, and real learning.
A History of AI Tools For Reading Comprehension
The story of language models begins with rule-based programming. In the early days of AI, machines were taught to identify objects through hard-coded logic. For example, to recognize a cat, a program might be given a series of statements: if an animal has four legs and says “meow,” then it must be a cat, right? While this approach was serviceable for basic tasks, it lacked nuance and adaptability.
The arrival of neural networks and deep learning changed this approach. By training on massive datasets of text, images, and audio, modern models can extract complex patterns and simulate reasoning in ways that resemble how humans think. Today’s LLMs, such as ChatGPT, can draw on vast fields of information and produce coherent, context-sensitive text in multiple languages. They’ve moved from passive tools to active conversational partners.
This article is based on a workshop from the Rethinking Reading: AI for Literacy Achievement workshop series, a set of webinars on education AI applications organized in collaboration with InnovateUS and the Burnes Center for Social Change at Northeastern University.
Interested in Learning More?
- Dive into this topic by viewing the full workshop.
- Explore more topics through other webinars featured in the Rethinking Reading series.
AI-Powered Personalized Reading Interventions
What makes language models especially compelling when it comes to reading instruction is their ability to personalize learning experiences through natural language interfaces. Teachers and students can interact with these systems conversationally, asking questions, requesting clarification, or generating practice prompts on demand.
These models, functioning as AI tools for reading comprehension, excel at adapting instruction to meet students where they are developmentally, making them ideal tools for personalizing reading comprehension. Through interfaces like GPT Builder, educators can configure personalized reading tutors tailored to students’ skill levels, interests, and learning goals. This creates new and easy opportunities for teachers to create lessons and classroom materials that students will find engaging and relevant to their interests. Using the tool, teachers can opt to emphasize vocabulary development, inference, summarization, or argumentative reasoning. Even the LLM’s tone can be tailored – warm and encouraging for a struggling reader, more rigorous for an advanced learner.
Multilingual support is another advantage. Because many LLMs are trained on text from around the world, they can operate in multiple languages, including many lesser-known languages that most teachers will not be familiar with. As AI tools for reading comprehension become more multilingual, they can support inclusive literacy strategies for diverse student populations.
Through interfaces like GPT Builder, educators can configure personalized reading tutors tailored to students’ skill levels, interests, and learning goals. This creates new and easy opportunities for teachers to create lessons and classroom materials that students will find engaging and relevant to their interests.
Best Practices: Implementation Grounded In The Science Of Learning
Despite their capabilities, LLMs are not plug-and-play solutions. Successful implementation requires alignment with the science of learning and a clear understanding of how students benefit from personalizing reading comprehension strategies.
First, the technology should augment, not replace, teacher expertise. Educators can use AI to streamline planning and generate instructional materials, but they should remain in control. As demonstrated in tools like GPT Builder, teachers can input specific learning profiles, curricular goals, and instructional priorities to guide the AI’s output.
Second, collaboration is key. Human-AI partnerships – where teachers edit and refine AI-generated content – yield better results than relying on automation alone. This ensures that prompts are culturally relevant, pedagogically sound, and tailored to the unique dynamics of each classroom.
Third, thoughtful design matters. Tools that tend to have the greatest impact are those that integrate social-emotional learning, accommodate different learning levels, and promote student agency. The shift toward personalization should be seen not just as a technical feat but as a teaching opportunity.
Human-AI partnerships – where teachers edit and refine AI-generated content – yield better results than relying on automation alone. This ensures that prompts are culturally relevant, pedagogically sound, and tailored to the unique dynamics of each classroom.
Personalized Learning With AI: Platforms To Consider
How can AI tools for reading comprehension be used to personalize reading interventions for students? Several platforms are already leveraging LLMs to bring individualized reading instruction into classrooms.
Amira Learning is an AI-powered reading tutor that originally focused on oral fluency through speech recognition. Its latest version leverages language models to engage students in real-time discussions, assess their understanding of texts, and adapt follow-up questions accordingly. Research cited by the platform shows significant gains in early literacy, well beyond the typical pace of progress.
Project Read takes a different approach, allowing teachers to upload curricular materials and automatically generate personalized stories that align with classroom goals. The system can insert a student’s name, tailor vocabulary to their skill level, and create content relevant to their interests. It’s a powerful time-saver for teachers and a motivating experience for students.
Luca AI is more student-facing, offering an immersive reading experience driven by interaction with an AI character. Students share personal details such as age and interests, which the system uses to recommend texts and create personalized learning plans.
Khanmigo, developed by Khan Academy, provides a suite of tools that combine personalization with pedagogical best practices. It’s “Make It Relevant” feature connects lesson content to students’ interests. A “Text Leveler” adjusts reading complexity for different learners. Other tools generate targeted discussion prompts and comprehension questions based on specific content or topics.
The future of AI-powered reading instruction depends on thoughtful implementation, ongoing research, and equitable access. As new tools emerge – including systems that support spoken as well as written language – educators must remain at the center of the conversation.
Challenges And Considerations
As AI tools for reading comprehension gain traction in schools, it’s important to critically examine their limitations and risks. Despite their strengths, language models are not without limitations. Most notably, they reflect the biases present in their training data. If not carefully monitored, LLMs can perpetuate stereotypes, generate inappropriate content, or overlook the needs of historically marginalized learners. These concerns are especially critical when personalizing reading comprehension, where sensitive content and cultural relevance matter deeply.
Access is another concern. Many of these tools require high-speed internet and powerful devices, which may not be available in all schools. While LLMs offer greater personalization than ever before, they also risk widening the digital divide if access is uneven.
Finally, transparency is an ongoing challenge. These models operate as “black boxes,” making decisions based on millions of variables. It can be difficult to trace how or why a particular output was generated. This complicates accountability, especially if a student receives incorrect or harmful information.
As AI tools for reading comprehension gain traction in schools, it's important to critically examine their limitations and risks. Despite their strengths, language models are not without limitations. Most notably, they reflect the biases present in their training data. If not carefully monitored, LLMs can perpetuate stereotypes, generate inappropriate content, or overlook the needs of historically marginalized learners.
Looking Ahead: A Tool, Not A Replacement
If used wisely, AI tools for reading comprehension have the potential to transform how students engage with texts and develop literacy skills. But they are not magic wands. Their power lies in augmenting what educators already do well: guiding, supporting, and inspiring learners.
The future of AI-powered reading instruction depends on thoughtful implementation, ongoing research, and equitable access. As new tools emerge – including systems that support spoken as well as written language – educators must remain at the center of the conversation.
Ultimately, language models are just that: models. They don’t teach on their own. But in the hands of skilled educators, they can help more students see themselves as readers, thinkers, and learners.
