The future of learning, if not overwhelmingly and obviously the near future of higher education, is in the powerful mix of adaptive, competency-based education (CBE), and gamification. (If you’re interested in exploring this in greater detail, Ryan Craig presents a very readableargument for this in College Disrupted.)
In this extensive article I introduce some basic concepts related to adaptive learning, provide some pointers to technology providers in the industry, describe the basics of how to integrate adaptive learning into a learning experience, describe some challenges that come when implementing adaptive learning experiences, describe some adaptive learning projects that we have implemented, and describe roll-out opportunities for universities and colleges looking to use this technology. Let’s get to it.
Some foundational concepts
What adaptive learning is and is not
Here is a definition from the Office of Educational Technology of the U.S. Department of Education:
Digital learning systems are considered adaptive when they can dynamically change to suit the learning in response to information collected during the course of learning rather than on the basis of preexisting information such as a learner’s gender, age, or achievement test score.
This isn’t what has been known as personalized learning in which paths are pre-defined for different learner types. This is far beyond that. This is a path that is built by an algorithm, usually in the cloud, based on data gathered for this user and integrated with data gathered from the successes and failures of every other user before that.
Another concept that adaptive learning gets lumped in with and confused with is competency-based education (CBE). This learning technology is about defining a learning experience (e.g., a class) in terms of the concepts that are learned and not the time spent in class (i.e., credits). Adaptive learning can be used to ensure that students effectively and efficiently learn competencies but it is not a necessary complement. These are both very important technologies for the future of education but they are not the same thing.
Components of an adaptive system
An adaptive learning system is composed of three separate parts:
- Content model: The system must contain a network or flow diagram of competencies and topics
- Learner model: The system must be able to represent the knowledge and abilities of each learner
- Instructional model: The system must be able to select content for a specific learner at a specific time
In addition to these core pieces, the effectiveness of a deployed adaptive system is also dependent on the following supporting systems:
- Student performance analytics: The system should provide an analytics dashboard that allows the teacher to determine how students are progressing, where they have been performing well, where they have been having trouble, and how the overall group has been performing.
- Content resource analytics: The system should also provide an analytics dashboard that allows the teacher to determine which resources are effective at teaching students. This is of great help when determining where and how to focus efforts on updating a class.
Types of adaptive systems
At least two different ways exist of classifying an adaptive system:
- Rule-based vs. Algorithm-based: Generally, earlier adaptive systems, or systems that are designed to be used on less expansive topics, use a rule-based system in which the content specialist is required to write a series of if/then rules that specify how the network of concepts in the knowledge base are related. The alternative is an algorithm-based system that largely relies on machine learning to create and update the network in the instructional and learner models.
- Content dependency: Adaptive systems also differ based on how much the three models and two analytics dashboards depend on the content on which it operates. Some systems are closely crafted and dependent on the content (e.g., it might rely heavily on the fact that it is an English language learning program or an introductory biology class); others might be completely independent and claim that they can perform well no matter what content is used; finally, others might claim that they are useful for mathematics classes in general but would not be useful outside of that field.
Technology providers
Adaptive learning systems have been and are being deployed in three major formats:
- Publisher: Many systems, most from textbook publishers, are fully integrated with specific content and are not obviously sold as or made available as adaptive systems. They are simply the supplementary materials (tests, quizzes, tutoring materials, etc.) accompanying a certain textbook that happen to be adaptive. This has essentially risen to the level of table stakes for the major textbook publishers.
- Platform: Some systems are general-purpose or targeted analytic platforms that include the three models and two dashboards described above. The user is empowered via supporting content creation tools to create his/her own analytic learning course.
- Embeddable engine: Finally, some systems can be used as an embedded analytics engine within a course platform. It would rely on systems integrators to use APIs to send data to and retrieve data from the engine. The integrators would then provide an interface and appropriate display for interacting with the learning resources and analytics dashboards.
It is beyond the scope of this article to go into depth on this but some leading companies (other than publishers) for providing analytics software are the following: Acrobatiq, Cerego, Knewton, LoudCloud, and Smart Sparrow. Each of these provides an interesting set of tools and supports adaptive learning in their own unique ways. Take the time to explore these and you will get a better idea of what is currently possible.
Integrating adaptive learning
If an organization, professor, or teacher is thinking about integrating an adaptive learning approach into a learning experience, several different options have been used:
- Fully adaptive paths: The entire learning experience and the whole knowledge map is fully available to the adaptive engine.
- Linear but adaptive within competencies: In this scenario the teacher sets up a series of major topics or competencies that will be covered in order. Within each item in that sequence, a knowledge map is defined upon which the adaptive engine operates. This provides the teacher with a mid-point between the traditional time-based, highly controlled course with which most faculty are familiar and the fully adaptive learning experience.
- Adaptive but with integrated human gatekeepers: This approach might be used when the student is expected to learn background, fairly straight-forward knowledge (which would be under control of an adaptive system) but then the ability to progress to the next topic or competency is under control of some human gatekeeper. This gatekeeper might assess a student based on a project, a discussion, a presentation, or some other subtle, nuanced method.
Other methods are possible, but the above demonstrate the variety of ways that adaptive can be integrated into a learning experience. It can be an extensive project or it can be a smaller, targeted effort that allows an organization, professor, or teacher to become more comfortable with the approach and build up expertise over time.
Challenges
While we here at Extension Engine think that adaptive learning systems are great, both an effective and efficient approach to learning, specific challenges definitely exist. Term lengths are the major impediment to the extensive implementation of adaptive learning across the board at a university or college. For the faculty member, an adaptive system can force them to address the question: “How am I going to organize my class and ensure that everyone does a sufficient amount of work if some of the students might be able to progress through the material at a significantly faster or slower rate than other students?” Given accreditation requirements as well as a sense of fairness held by some faculty and students, the answer cannot simply be that it doesn’t matter or it won’t have that big of an effect. Some adaptive systems will allow a student to learn the material significantly faster than a traditional system. The faculty member must decide how to address this.
Another challenge with rule-based systems is, not surprisingly, writing the rules. While writing any one rule may not be that complex, writing a whole knowledge base of rules can quickly become overwhelming as the interaction between rules and topics can be difficult to track. This complexity is the primary reason that rule-based approaches to adaptive learning are most useful for smaller learning experiences.
The final challenge I discuss here is that adaptive learning systems are a new approach to teaching for many professors. Adaptive learning can take some of the direct control of the learning process out of the hands of the professor. It can put the professor into a new role, not as dispenser of all information but as a guide or mentor in the complexities or more challenging aspects of the field. It can be an exciting transformation, but it can be a definite change that must be addressed.
Future roll-out opportunities
Universities can deploy online learning — and, specifically, adaptive learning — in several different ways.
- Pre-matriculation: Online adaptive learning programs might be used to attract students to a school by teaching one topic in a popular subject at the school. Another possible use might be to deploy an adaptive learning experience that prepares students for their first year of school, either by requiring that each student complete an adaptive math, economics, or whatever course before they enroll. For some students this would not take that much effort while for others, their preparation for college (and probability for success) might be significantly raised.
- Post-graduation: A university or college might deploy an adaptive learning program as part of either an alumni engagement campaign. The school might create an adaptive course either covering the foundations of a field for a widely varying alumni audience or introducing recently discovered updates of knowledge. Another set of courses might be created to address the continuing education market. An adaptive course, again, could be deployed for a widely varying audience while still enabling it to cover a topic in sufficient breadth and depth for each individual student.
- For credit courses: This was discussed above in the section “Integrating adaptive learning”.
- Marketing & research (e.g., MOOCs): This might be the most attractive and appropriate place for the use of adaptive learning technologies. Given the potential size of the audience, the variety of their background and preparation is almost certainly quite wide. Further, given that a MOOC has generally been taught outside of the constraints of a degree-granting context, its use has been experimental for both the organization and the professor. This is a perfect context for getting comfortable with the adaptive learning approach.
Conclusion
For almost any university or college, an appropriate opportunity exists for deploying an adaptive learning experience. Adaptive technologies at any scale are available and ready to use. The tools market is still in the early stages so winners and losers cannot yet be identified; however, the tools are at a sufficient maturity that they can be useful for organizations, professors, and students.
Adaptive learning will soon be table stakes for higher ed. Basically every K-12 book is or will soon use adaptive learning in supplementary (or even primary) material when it is online. College students will soon come to expect this approach in their learning experiences.
Administration and faculty should be gaining experience with this approach as quickly as possible. If they feel it is too risky to deploy this approach in for-credit courses, then they should use the other approaches (described above) to explore and experiment.
Keep Learning
Define and Act on Your Institution’s Strategy
A 15-part series on defining and acting on a higher education strategy to guide leaders during these difficult times. It is targeted at educational leaders who are participating in shaping their school's actions during and after the COVID-19 pandemic.
