How To Use AI For Teaching Reading and Writing

By Kristen DiCerboPh.D., Chief Learning Officer at Khan Academy

Reading and writing are so much more than skills for the English classroom; rather they are keys to critically interacting with and changing the world. In order to ensure that students have the skills to navigate an increasingly-complex and information-rich world, we are increasing our literacy offerings, all designed based on the principles of learning science.

Developing deep reading comprehension

As students approach 4th grade in the U.S., an important shift occurs in their education: the emphasis moves from learning to read to reading to learn. This means that students are expected to extract new information from text, interpret it, remember it, evaluate it, and recall it when needed. This shift to reading to learn marks a pivotal turning point in education and if students are not able to make the shift, they will fall further and further behind across subject areas.

In order to proficiently read to learn, students need to master a series of individual skills and then synthesize them to construct an integrated understanding of a text. In order to ensure that students are able to do just that, we recently released a new 6th grade reading and vocabulary course that includes both directed practice in individual reading skills, such as analyzing text structure, identifying point of view, and tracing ideas across a text, as well as integrated practice designed to allow students to build a broader and deeper understanding of extended passages.

In addition to the challenges posed by mastering reading skills, students may also face barriers to reading comprehension if they lack vocabulary and background knowledge that is integral to the text. This was clearly pointed out to me with an example from sports: I know a lot about baseball and, given a piece of writing about baseball, I can read and remember it with no problem. However, give me a passage about cricket and I am lost. I don’t have the necessary vocabulary, so I can’t get meaning from the passage overall. I don’t have any prior knowledge to connect the information in the passage to in order to help me remember it.

To alleviate these challenges for students with different funds of knowledge and to ensure that each learner can come to each unit with an adequate knowledge base, each unit in the new course begins with a lesson called “Building Background Knowledge.” This lesson provides learners with access to useful academic vocabulary that students can expect to see not only within the course, but in their reading across the curriculum. The lesson also includes an overview of the unit’s content, as well as essential questions to focus students’ thinking as they work through the thematically-linked passages they will encounter across the unit.

And because we recognize the significant role that academic vocabulary plays in reading proficiency over time, each unit also ends with a unit vocabulary review and a quiz. The quiz represents a key mechanism for transferring acquired vocabulary from short- to long- term memory,  because it employs the well-researched principle of retrieval practice. The practice of retrieving something from memory once it is learned is a vital means of improving learning outcomes, and low stakes quizzes are an excellent way to do this.

In addition to our course activities, we have a variety of playful AI-powered activities to support students’ comprehension. For example, to practice vocabulary skills, students can play Word Safari and Word Architect, where learners interact with Khanmigo to understand word construction and meaning in engaging ways. And to continue building background knowledge, students can use Khanmigo to “Chat with a historical figure” to engage in discussions about events and perspectives of the past.

Mastering writing

We want our learners to not just consume text but to also be able to express themselves in writing. Becoming a good writer, like any other skill, requires both frequent practice and access to meaningful feedback from skilled practitioners. The National Commission on Writing recommends that students spend 60 minutes per day writing, yet data from the National Assessment of Educational Progress, as analyzed by the Learning Labsuggests that only 31% of high schoolers are writing even 30 minutes per day.

The report paints a vivid picture of the challenges of writing instruction. Only 13% of teachers reported asking their students to write persuasively weekly, and 23% assign a paragraph or more of writing every two weeks or less. Such limited opportunities to receive meaningful feedback on their writing can represent an enormous barrier to students’ development as writers. Yet we can’t fault teachers for this lack of feedback: 50% of teachers already spend four or more hours per week grading, an already-unsustainable practice. It’s an unfortunate fact that giving feedback on writing is a huge time commitment for teachers. So how can we get students to write more without requiring even more time from teachers?

Historically at Khan Academy, we have not offered support for teaching writing because we did not have a method to provide meaningful feedback to learners. We know that a key benefit that educational technology has provided in some domains (particularly math) has been the ability to provide individual learners with immediate feedback. However, until recently, we had not been able to replicate this in writing, where feedback occurs in layers ranging from the thesis statement to the overall structure and on down to grammar. However, recent advances in generative AI have opened up possibilities for new and better writing feedback.

Our recently published Khanmigo activity “Give feedback on my academic essay” allows learners to submit an essay for feedback. Learners get feedback on the introduction, evidence, structure, conclusion, and style. In order to do this, we have spent eight months perfecting prompts for each element, in order to teach the AI model how to give feedback for that aspect. For example, when evaluating an introduction, Khanmigo examines whether there is a clear argument or thesis statement and whether a student author has provided enough context for the reader to understand what the essay is about. In addition, rather than just providing the feedback to learners and hoping they do something with it, the tool allows students to ask Khanmigo questions about the feedback, make revisions, and get immediate feedback on those changes.

Integrating technology and ELA

We are excited about the coming possibilities for ethically and rigorously integrating AI with the teaching of English Language Arts. Technology support for ELA has long lagged behind math and science, because we haven’t had the tools to provide more qualitative feedback and in-depth interaction. Thanks to the opportunities provided by AI, new possibilities are arising every day. We look forward to working with all of you to see what we can do to improve reading and writing outcomes for learners.

To learn more about bringing Khanmigo tools to your students, visit https://districts.khanacademy.org/khanmigo