Can AI be your best teacher?
In the previous blog, I addressed how ChatGPT doesn't mean the end of learning and highlighted the importance of teachers continuing to instruct fundamental skills like remembering and understanding. These foundational skills serve as the basis for students to acquire and master complex skills that demand higher-order thinking, such as evaluating or creativity. In this blog, I am wondering whether AI can be utilized to improve learning outcomes and if an AI system could potentially even be your best teacher.
The different skill levels
To understand this, let's break down a teacher's role into two broad categories. The first might be considered that of a taskmaster, responsible for imparting facts and skills. The other role is that of a motivator, guiding students, assisting in formulating life goals, and inspiring them to achieve those goals. The former role necessitates a wealth of subject content knowledge, while the latter requires strong socio-emotional skills. Each teacher assumes these roles to varying degrees based on the student and their needs. In my view, an outstanding teacher is one who strikes the right balance between these two roles. Let's consider how these roles align within the Bloom's Taxonomy framework, where skills progress from simple remembering to the highest level of creating. Each level builds upon the ones preceding it.
Mastering lower-order skills requires extensive repetition, timely feedback from the teacher, and further practice. Only through this iterative process can students truly master these fundamental skills, enabling them to progress towards higher order skills like evaluation and creation. These higher order skills depend heavily on background knowledge and basic skills learnt through repetition. However, spaced repetition is not inherently engaging, inspiring, or perceived as exciting. It can be monotonous to repeatedly work on basic grammar, draw the reflection and refraction of light rays, or practice basic algebra. Despite this, teachers must act as taskmasters, encouraging students to persist in their efforts.
During these cycles of repetition, students inevitably encounter obstacles, and these challenges vary among different students at different times. Unfortunately, teachers often lack the resources to be present for every student at every moment of need, resulting in potential gaps or "holes" in students' foundational knowledge. These holes in knowledge and skills make it more challenging for them to master more advanced skills. Consequently, many students may struggle to reach their full potential due to this lack of a strong foundation. In this context, I believe AI can offer a promising solution. Before I discuss that, let us see how assessment and feedback is related to skill levels.
Assessing skills: from specific feedback to guidance
Lower-order skills are generally easier to assess and provide specific feedback on due to their definitive nature. For instance, it's simpler to evaluate the grammatical correctness of a sentence, which represents a lower-order skill, than to assess the quality of a creative essay, which constitutes a higher-order skill. Similarly, identifying mistakes in a ray diagram in physics, a lower-order skill, is more straightforward compared to grading a group project, which represents a higher-order skill. The distinction between right and wrong becomes less clear as the complexity of the required skill level increases. Evaluating higher-level tasks often becomes more subjective, involving individual judgment and open interpretation. Consequently, the assessment process shifts away from seeking definitive correctness and leans more towards guidance and mentorship, as specific feedback becomes less meaningful in the face of subjective evaluation.
So, how can AI help? Let us consider the example of a student learning algebra by interacting with an AI system. It could, for instance, be through a game based system like ViLLE or Brilliant.org. The game like environment means that they already have instant feedback features. What if they can be also be powered with AI natural language capabilities, like Khanmigo? What if this system also has the capacity to generate personalised learning paths on the go? To take it a step further, could it even be a virtual reality game that the student steps into when it is time to learn? The student could complete one challenge at a time with an AI avatar offering assistance as the student progresses through the game.
Ideally this is what a teacher does within the four walls of a classroom: giving feedback and creating personalised learning plans. But the feedback is not always timely nor are the learning paths really personalised. However, unlike a teacher, the AI is always present. It is always available. It can give well timed help as per the needs of the student. This eliminates the need for a student to feel awkward about asking questions in front of peers, fearing embarrassment. Instead, the student can privately interact with an AI model (tutor) without the social context hindering their learning experience. Additionally, the AI can continuously monitor and assist without the teacher having to identify struggling students while circulating the class. It offers a more personalized and proactive approach to support.
But learning is not just about knowing facts or mastering basic skills. We want our students to be creative thinkers. We want them to be able to tackle complex problems that world is facing now and will face in the future by working together with others. These are higher order skills. Skills that require not just content knowledge but also strong socio-emotional understanding.
As the student climbs the ladder of skill complexity, she can then start working on these creative tasks. These tasks do not necessarily have a clear right or wrong. Instead of specific feedback, they require mentorship and guidance. They require working collaboratively with others and ability to self-reflect. This is where humans can play a pivotal role, fostering compassion, encouraging collaboration, and helping students develop their socio-emotional and self-reflective skills. For skills that fall in the mid-range of Bloom's Taxonomy (apply, analyze), the teacher could work collaboratively with the AI. This idea is summarised in the figure below. As Perttu Pölönen nicely summarises in his book, 'Tulevaisuuden lukujärjestys', thanks to technology, we might be slowly making our way from our heads to our hearts.
In fact, this is an aspect of training I wish I had received more of - the simple essence of being kind and humane. The Finnish teacher training focussed heavily on preparing pedagogically sound teachers with ability to develop right teaching methods for each topic. There was also emphasis on self-reflection and continuous improvement, maybe indirectly aiming for humanness qualities. But looking back, I wish I had more explicit training on topics of motivation, leadership, compassion, fairness and especially how they apply for students with special needs. The need for special support for students is becoming increasingly relevant in Finland and the support system is not working as one would have hoped for. With the right training and resources, we teachers could be more effective at providing support to these students, taking weight off the system.
Also, the effect of AI in education can potentially be tremendous especially in regions around the world struggling to equip students with basic skills. According to UNESCO press release, there is a huge need for qualified teachers in developing countries, particularly in sub-Sharan Africa. The teachers on service are also unfortunately plagued by poor content knowledge and high degree of teacher absentism. Training teachers require huge infrastructure and we humans are particularly slow at learning new things. According to OECD report, AI is predicted to solve all literary and numeracy tasks by 2026 while average human literacy and numeracy levels have stayed roughly the same over the past decades. Therefore, it seems to be just a matter of time before AI outperforms teachers in terms of content knowledge. In fact, the potential of AI to address several of United Nations Sustainable Development Goals (SDGs) in Africa is already recognised. Could AI help tackle SDG 4, to provide quality universal basic education, considering how we are already lagging considerably behind in this area?
What is the verdit?
To finally answer my original question, "Can AI be your best teacher?", my answer would still be a no even though AI could surpass most teachers on content knowledge.
My most cherished memories of my teachers are from moments when they courageously revealed their human side—when they listened to me with compassion, understood me, and motivated me. I am confident that the same holds true for most students who have had the privilege of being taught by exceptional teachers. They often say that good teachers "truly care." As a teacher, my most memorable moments have been when I found the time to care for the students who needed it.
I readily acknowledge how technology is better than me in various subject matters when I use resources like YouTube videos or ViLLE exercises to impart content knowledge. I personally would not have a problem giving up more of this space to a pedagogically sound AI if it means that I have more time to get to know my students and to really care.
So, could we rethink what teachers do? Can teacher training focus more on developing human skills? Could teachers be trained to work alongside AI systems that handle the content knowledge? These are not easy problems, and we have a lot at stake. This requires very careful thinking, especially concerning data and student privacy. There are perhaps lessons that we can learn from China (both good and bad) from their head start in AI in education. What we should not hopefully do is shut down the possible use of AI in education out of fear of the unknown. We do not want to throw the baby out with the bath water. As Sal Khan says in this video, AI can be used to enhance human intelligence, human potential, and human purpose. Can AI make us better humans? I don’t know. These are fascinating times for the field of education and humanity in general.
