AI and Education's Eternally Unfulfilled Promises

Education has faced, for a long time already, a state of affairs of unbearable perplexity: despite incessant proclamations, speeches, books and even innumerable professional development programs that outline a clear future for learning and schools, we fail to act on them, and we remain rooted in pedagogies and practices that were effective... for a world that does not exist anymore.

Of the many unfulfilled promises that we educators are guilty, 3 stand out, not only because of how far we are from making them a reality, but also because they are quintessential to a renewed vision for education.

Personalizing Learning

We have long known that the infamous one-size fits all is a sure recipe for failure, and demanding the same of every student has led to a sad legacy of irreparably damaged collective self-esteem for many generations, whose effects are evident only years later. Both instruction and assessment have prioritized standards and outcomes, both terms that have been long out phased in the real world. Even though there is universal consensus to the effect that it is essential that we cater to students′ needs and customize learning, personalization remains, still, a largely utopian proposition.

It is rightfully argued that teachers face impossible demands and that personalizing contents and assessments for more than 15 or 20 students is a daunting task, and, regretfully, this practical hurdle has impeded the materialization of the vision of personalization in education, to the extent that the powers to be, and even the general public, are reticent to follow up on the demand for personalization.

AI can clearly come to the rescue, in what is clearly one of the effective applications of AI for learning:

• Adaptive systems. These online learning systems that are based on big data and AI start from diagnostic testing of students and then provide an optimal learning path through comparing learner profile and interactions with the many others that it has stored in its database. In the same way that an e-commerce portal can predict with uncanny ability what we want to buy without us thinking of it, these systems capitalize on the powerful combination of AI and Big Data to truly personalize the learning experience for each student.
• Personalized tutor. A unique and unprecedented feature of AI systems is that they can act as an interactive, infinitely patient, resourceful and personalized tutor who can explain, as many times as needed, any topics, provide examples, guidance, and help students navigate even the most complex topics by providing ample learning support.
• Inventory of students’ interests and passions. By having students complete extensive surveys and then processing them through AI, it is easier than ever to create a veritable catalog of students’ interests and passions, for each one of them, for future reference in the creation of learning materials
• Developing personalized learning materials. One of the main strengths of ChatGPT and a myriad of other generative AI applications is the ability to generate customized learning materials, from reading lists, to articles, problems, case studies, and any content related resources that can personalize the learning experience for each student. This can be applied to the whole range of subject matters, from foreign languages to math and science, as well as providing invaluable help in creating interdisciplinary projects or linking with real life.
• Personalized assessment instruments. Similarly, GenAI can very easily create personalized assessments that cater to students′ learning styles, strengths, and even areas that they need to improve on, to any level of difficulty.

Learning from Mistakes

Another flagrant disconnect for education is the much sought after objective of learning from errors, creating a safe environment where students are unafraid to make mistakes and develop the resilience and resourcefulness needed to overcome setbacks in real life.

Inasmuch as learning from our mistakes has become a kind of universal societal mantra for success, the harsh reality of schools is markedly different. Averages and our perennial obsession with standardized measurements are completely contradictory with this objective. I have yet to find a student who receives a low grade and subsequently looks upon it as an opportunity to learn the value of resilience.

The greatest tragedy of school is not that students do badly, but that they would much rather be passive learners than being stigmatized by their peers, and, sadly, not infrequently, by the adults themselves, when they make mistakes.

AI can, once more, become a game changer in this respect.

• Learning through games and simulations. Gamification, as we long known, provides the ideal scenario for learning in that it provides an engaging, fun environment with real-time feedback that helps students learn by trial and error. A teenager does not give up and cry foul in frustration when they are, literally, killed in a game, but rather persist, until they can muster even advanced levels. The application of AI to learning games can provide the ideal environment for collaborative real-time learning where students are not only not stigmatized but positively challenged by their failures.
• A safe environment for learning. The importance of a safe, aseptic environment where students are unafraid to make mistakes cannot be emphasized enough. In the same way that there are studies that, amazingly, seem to indicate that humans prefer online therapists to real persons, interacting with an AI for learning, as we saw in the previous section, takes away the element of shame in not understanding, and asking for as many explanations as needed to a level of detail that come be catered to each learner.
• Formative assessment. Long hailed as the definitive evolution of assessment, assessment for learning, where students are given feedback that lets them improve their productions and are not punished by their mistakes, is still mostly wishful thinking, despite it being the most talked about topic in education. The paid version of ChatGPT, for example, can already provide personalized detailed feedback for any written assignment utilizing a given rubric and tapping into its extensive database to give students suggestions for improvement to a level of detail and precision that are impossible for any human. If teachers judiciously utilize this feature, they can then better use their time to discuss, in person, that AI-generated feedback with each and every student, in what is, ironically, shift towards a more human interaction with their students than just grading a paper.

Inclusion

Perhaps the greatest, and saddest of all our inconsistencies, is our inability to include students who require special focus from the teacher, and/or have emotional issues or learning difficulties. Once more, despite the prevalent discourse and even well-intentioned efforts to provide for students that don’t fit the archaic concept of normality, inclusion is far from the reality in most schools, even the best resourced.

The real-life challenges are akin to those of personalization, since inclusion is, ultimately, a targeted effort in personalization, albeit if in a scenario that requires a lot more energy and attention. When teachers have to provide for the whole class, it is unfeasible for them to engage faithfully in the infamous “adaptations” for students to be able to access learning.

In this respect, once more, AI can be a truly invaluable assistant, since, in particular, ChatGPT “knows” about most learning difficulties, emotional issues and mental illnesses and can adopt, customize, and even produce completely new learning resources that are specifically targeted for each student.

The paid version can even scan through a whole document with the learner′s unique profile and incorporate that into a custom GPT that can be subsequently invoked to personalize the learning.

This also includes, of course, not only developing learning materials but also assessments that can allow students to demonstrate their learning and receive positive feedback that allows them to continue in their growth process.

As we all know, there are many other features of the promised land of the future of learning to which we have only paid lip service, such as multimodal assessments, fostering creativity and higher-order thinking skills, relating with real life, and learning by doing, with as much connection to real life as possible.

In all of these cases, and, as we have said in previous articles, irrespective of any philosophical discussions or controversies, GenAI can be the definitive and powerful catalyst that allows us to make true on a vision for the future of learning that has not changed, but is much more within our grasp.