Most K-12 schools in the US will continue to use learning management systems (LMSs) to deliver learning materials, including documents, videos, and quizzes.1 Look one level deeper, however, and the way schools use their LMSs will undergo shifts that have big implications for whether their edtech has a transformative impact.
In recent years, educators who want to provide digital content to their students choose among three instructional models: the teacher-directed instructional model, the software-directed instructional model, or a hybrid of the two.2 For reasons I explain below, the software-directed instructional model is superior when netted out, but most schools won’t use it next year. This will cause the sliver of schools that use the software-directed method—including microschools, homeschools, and other entrants—to pull ahead dramatically in student academic performance in 2025.
Defining the three instructional models
Schools have three options for how to provide digital instruction on their LMSs.
Teacher-directed Instructional Model
In this model, teachers create the curriculum themselves. They begin by identifying learning objectives based on the standards they want to teach. Then, they find or create digital content, objective tests, and performance tasks aligned to those learning objectives, which they post to an LMS.3
One benefit of the teacher-directed instructional model is that teachers get complete control over how the content is taught (they can even video record themselves teaching it). Many teachers enjoy how this model allows them to prepare and post their course content, and students may prefer having their own teacher instructing them.
The main drawback is that this model simultaneously holds all students to the same learning objective, preventing them from getting the proper lesson unless the teacher has built thousands of lessons and allowed students to jump back or forward to different grade levels according to their levels. Also, only some teachers have the time or money to invest in building courses with the functionality, attention to the user experience, and variety of hints and examples that a third-party software vendor can afford. Finally, if a teacher isn’t an engaging presenter, the lecture is likely even worse on video than in person.
Software-directed Instructional Model
In this model, teachers guide their students as they progress through third-party learning software, such as iXL, Math Academy, Khan Academy, or i-Ready.4 Assuming the software has been well-vetted already, then the basic workflow for teachers is two steps:
- Establish a minimum pacing standard. For example, students must complete X lessons, earn X game points, log X “time on task” minutes, or master X skill assessments.
- Support students as they drive their own learning.
The benefits of the software-directed instructional model include removing the burden on teachers to build their own courses, thereby affording them more time to guide their students individually. This model also offers students more goal-setting opportunities to nurture their ability to take responsibility for their learning.
Those benefits aside, this model’s real game-changer is that it allows students to learn at the right level. AI-powered adaptive learning software can analyze students’ performance, strengths, and weaknesses to provide personalized learning pathways tailored to their specific needs and abilities. By adapting content difficulty and pacing in real-time, these systems ensure that each student is consistently challenged at an appropriate level. Learning at the right level is one of the most powerful solutions our world can give children, as I argue here.
Drawbacks to this model include that some teachers or districts don’t want to give up the opportunity to deliver their own lessons, the software can be poor quality and poorly designed if not well-selected, and appropriate software does not exist for every subject and course.
Hybrid Instructional Model
The hybrid instructional model combines elements of both the teacher-directed and software-directed instructional models. Teachers use self-created content and third-party learning software to post instruction on their LMS.
The benefits of a hybrid approach include that teachers spare themselves from recording every lesson; they can rely on Sal Khan and others to do some of that work. Also, students get to see their teachers’ familiar faces and get some time to learn at the right level.
The glaring drawback is that the teacher’s self-made content in this equation forces students to work on the class’s level instead of their personalized level. Suppose Ella is working on 9th-grade English as a 10-year-old. The content her teacher makes for the LMS is likely several grade levels beneath Ella’s level and a waste of her time. She’d be better off with entirely software-directed instruction, not a hybrid.
Predictions and their justifications for 2025
Most teachers will continue to do the teacher-directed or hybrid model in 2025. More than three-quarters (77%) of teachers who relied on curriculum outside of textbooks created their own materials in 2023, and that pattern is likely to continue. In fact, there was a 5‒6% drop in teachers who turned to commercial products for non-textbook instruction in 2022‒23 versus 2021‒22.5
The theory of hybrids, a companion theory to Professor Clayton Christensen’s theory of disruptive innovation, justifies this prediction. The theory of hybrids says that whenever a disruptive technology emerges, the leading firms in the field typically want to use it. They perceive it as not yet good enough for their customers, so they develop a hybrid. The hybrid technology marries the old technology with the new to create a “best of both worlds” combination.
In like manner, most incumbent schools see the advantages of AI-powered adaptive software but shy away from completely replacing teacher-directed instruction. Instead, teachers build a hybrid solution on their LMSs.
Meanwhile, entrant schools outside the incumbent system are seeing the value of the pureplay disruption—AI-powered adaptive software—and will increasingly use it as the primary, if not exclusive, source of instruction in 2025. Rising examples include Acton Academy, 2HR Learning schools, Astra Nova School, Prenda, and new district-based alternative schools. Anticipated improvements in AI in 2025 only strengthen this prediction.
Expect these pure plays to get significantly better results compared to incumbents, assuming the other pillars of their schools, such as the on-site guides, organizational cultures, and custodial care, are adequate. Learning at the right level in a way that replicates the effects of human tutoring has proven in innumerable studies to be a more effective instructional design than monolithic, one-sized instruction.6
Bright possibilities
The bad news is that roughly 94% of students attend more or less conventional classrooms in the incumbent system.7 That implies most students won’t benefit from pureplay software-directed instruction in 2025.
But there are two bright spots. First, the number of students in startup schools is poised to grow as school-choice policies expand. State-funded education vouchers create a funding mechanism for startup schools to flourish.
Second, the incumbent system might be handed a win. Clever LMS providers are starting to find a way to overcome the disadvantages of teacher-directed and hybrid instructional models. They’re beginning to embed AI-based tutors directly within their platforms, which create personalized accommodations for students in real-time. For example, Microsoft Accelerators simplify and explain texts and math problems when students struggle with teacher-made source material. Agilix’s BusyBee “learning companion” provides customized strategies and hints on the Agilix LMS when it perceives students are struggling with the teacher-made source material on the LMS.
Perhaps these LMS innovations will become so good that they can serve full courses of advanced and remedial lessons in real-time to supplant the underlying teacher-made content. If so, teachers in incumbent schools can move forward with their teacher-directed and hybrid instruction. Meanwhile, the LMS platforms will adjust that instruction to personalize it for each student. That moonshot possibility would be a win for those teachers and their students.
- Approximately 94% of K-12 school districts use at least one LMS, with only 6% of educators reporting that their district does not use an LMS. Lauraine Langreo, “What Teachers Really Think About Their Learning Management Systems,” EducationWeek, Sep. 22, 2022,
https://www.edweek.org/technology/what-teachers-really-think-about-their-learning-management-systems/2022/09
↩︎ - Miami-Dade Public Schools illuminated these categories for me in its publication “Instructional Continuity Plan 2.0,” Apr. 4, 2020, https://api.dadeschools.net/WMSFiles/224/icp-plan/MDCPS-ICP-PLAN.pdf. ↩︎
- Objective tests are usually multiple-choice quizzes or other tests with single correct answers. Performance tasks require students to apply their knowledge, such as through an essay, performance, work of art, exhibition, and so forth. ↩︎
- For more examples, see “30 Examples of AI-Adaptive Software,” Guide School, https://drive.google.com/file/d/1XsqsTQzb6S3cm8o_ZHsMhMinx06Jpqot/view?usp=sharing. Two helpful indexes for finding options are the EdSurge EdTech Product Index and CommonSense Education. ↩︎
- Julia E. Seaman and Jeff Seaman, “Curricula of Many Sources: Educational Resources in U.S. K-12 Education, 2023,” Bay View Analytics, 2023, https://www.bayviewanalytics.com/reports/curricula-of-many-sources-2023.pdf. ↩︎
- See, for example, Andre Joshua Nickow et al., “The Impressive Effects of Tutoring on PreK-12 Learning: A Systematic Review and Meta-Analysis of the Experimental Evidence,” Brown University, EdWorkingPaper No. 20-267, https://edworkingpapers.com/sites/default/files/ai20-267.pdf. ↩︎
- Assume most students outside of conventional classrooms are in homeschools or microschools. According to the National Center for Education Statistics, roughly 5 percent of children ages 5 to 17 were homeschooled in the US. According to Robert Enlow, CEO of EdChoice, roughly 1 percent are in microschools. That means roughly 6 percent of students are in these non-incumbent schools. See https://nces.ed.gov/whatsnew/press_releases/9_17_2024.asp and https://thehill.com/homenews/education/4730264-microschools-small-classroom-size-school-choice-movement/. ↩︎
