Algorithmic Education
Education has gone through a lot of changes in recent years. The main change is the use of technology, resulting in hyped terms such as elearning and EdTech. The main idea around using technology in education is to provide a learning experience that adapts to the learner's needs, preferences, and level of knowledge. A classic example is learning a language. Learners will have different knowledge levels, and technology is expected to pick this up, based on the learner's interaction with the system, resulting in different learning experiences. In theory, allowing learners to have a more adapted learning experience with the use of technology seems like a good idea. In practice, however, it has the classic limitation of cost. It is expensive to create digital learning content and even more expensive to create it for many different learners. This has led to mostly categorised learning paths, such as beginner, novice, advanced, etc., in which either the platform or the learner can select what group they belong to, with each group having a different learning experience which was pre designed.
The advancement of algorithmic learning, however, has shown some potential in creating more flexible learning experiences on the fly. Large Language Models (LLMs) for example are perfectly suited to serve as a learning mentor or learning buddy. Both paradigms learning mentor and learning buddy are well explored in serious games, a genre of games that nicely blends learning with gameplay. Large Language Models can, based on dialogue, explore the learner's knowledge level and can adapt educational challenges for the learner.
This is useful for small courses, short learning experiences, and specifically for topics that can be easily assessed, such as language classes or coding courses. However, how does algorithmic education that is education that is supported by technology support longer learning paths, such as high school and higher education? Meaning, can the concept of a learning buddy or mentor, both represented by an algorithm, have educational value, and does it support the already existing structures or does it create a more lonely and isolated experience which is already fuelled by automated and prerecorded online classes?
Learning as such is a more social experience is a path that algorithmic learning needs to consider and is not considered currently. A blended mix of supporting with short term learning tasks, for which LLMs are useful, with long term learning experiences in fast changing environments. Long term learning outcomes are often hard to measure and evolve around learning how to learn, how to socialise, how to adapt to new challenges, how to creatively find new solutions to problems unknown before, and so on. Algorithmic learning can in those cases be used as a motivator or guiding principle but cannot replace traditional structures such as parenting, friendship, and community.
Algorithmic Education might be the cornerstone of a bigger picture leading towards Artificial General Intelligence. It is hard to imagine, however, how Algorithmic Learning, how it is currently designed, can support learners with more complex and long term learning goals, for which traditional institutions are set up. It is plausible, however, that Algorithmic Education forms a support mechanism which can help the learner to stay on track and motivated, especially short term.
One final thought is to go all the way, to imagine a completely algorithmic university. This is however founded on the principles on which universities are founded: as beacons of change and even resistance in the face of commercial pressure and political interventions. Universities hold the light within that creates the next big thing, be it scientific or social in its breakthrough. For an algorithmic university to prevail, the convergence of algorithmic personal assistance technology merged with a larger goal of guiding educational experiences towards learning to be a critical and independent thinker. For algorithmic education to get to that point, the algorithm itself has to sit with the individual in a way that it can both support the individual but also empower the interfacing with others, that being other humans and their personal and supporting algorithms. We can therefore and have to be aware of algorithmic developments not resting in the interest of corporations, as they are tending to do, otherwise algorithmic education will not be able to lift itself onto the status of a free and critical thinking supporting institution such as a university.