AI, Polarisation, Qubits, and Gutenberg!

A range of LED bulbs, some lit to show the number 0 and some to show the number 1

The past few years witnessed a sharp rise in thought binarism in general, exacerbated by many political tensions, leading to a dichotomic approach in most things, including views of technological advances. When it comes to artificial intelligence (AI), for example, it feels like the norm nowadays is to adopt one of two extremes: you either love it and go all in, or you hate it and avoid it like the plague. While this applies to many fields of application of AI, the polarity is quite blatant in the context of AI in education. People promoting the use of AI in education believe it makes the lives of practitioners easier by taking on the heavy lifting; they trust that AI allows personalisation of learning; they also say that the positive outcomes outweigh the negative ones, and that when in doubt the user can do the sifting as needed; all of which are valid points. Those against AI argue that it might ‘steal’ the jobs of teachers, curriculum developers and assessment writers; they stress that the cognitive offloading with AI is harming learners; and they talk about biases, hallucinations, ethical and even environmental impacts, as well as a larger digital divide. These are certainly sound concerns. So, where should one stand?

Before talking about AI tools though, let us start with divides (yes, in plural form). A lot is being said about a (greater) digital divide being created with AI, and I personally worry about it seriously and wrote on it,¹ but I have a feeling that AI led to other types of divides that are not addressed as much, divides that are not helping with bridging the digital one: a divide in principles; a divide in visions; a divide in strategies; a divide in priorities, at a policymaking level, and even within communities of researchers and educators. What should we do next, now that the genie is out of the bottle and is not going back in? Of course, there is a lot of talk around a commercial bubble that could burst soon, but I am not talking about the companies and stock market here, but the technology and science behind it. Many feel that AI’s impact goes beyond how it affects educational resources or even teaching and learning overall; they say that AI use (or overuse) might eclipse the human in us, as in eroding our creativity, critical thinking, social interactions, emotional elements, and more. Others say that AI facilitates things and allows us more time to focus on being creative and innovative, so its adoption should be encouraged and introduced as early as possible. The question, in my opinion, should not be ‘Do we use AI or not and to what extent?’, but rather ‘What do learners need in this time and age and how can we achieve it?’. Whether AI turns out to be a facilitator or an obstacle, it is something that arises in the discussion and we can then study it carefully. AI should not be the focus while education becomes the side talk or simply a testing field. We must agree on our priorities first, and then discuss all the AI-related questions, logically and scientifically, to ensure we reach some common ground. In today’s education scene, it is not differences of opinion or approach that we witness, nor healthy disagreements and constructive debates, but rather a polarisation around AI, in how to use it, if at all; how to embed it in education, if at all; and whether, to put it simply, AI is good or bad. This is as binary as it gets. AI is bringing much more out of us as humans, not just as educators, than objective thoughts and genuine reflections, and the main focus points that deserve to be discussed get lost in the maelstrom of emotional debate. The focus is on regulation more than education; the focus is on technology more than pedagogy; the focus is on winning a personal fight rather than seeking what is right; the focus is on comments of the I-told-you-so type rather than thinking about learners and teachers beyond the hype. This has to change. Education, pedagogy, what is right and most useful, should take center stage, while the rest can be designed around them.

Back to AI resources; while I appreciate the excitement of many and I understand the scepticism of others, especially based on the use of large language models (LLMs), let us start by acknowledging that AI is not restricted to LLMs, and definitely not to a selected few of them. Therefore, taking a decisive stance for or against the concept of AI because one or a few tools perform well or do not always get it right (yet?) is not reasonable. It is like trying a new dish; you cannot judge the entire AI cuisine based on one good or bad experience! Moreover, academic research in the field is not conclusive enough at this early stage to back this or that theory; it provides food for thought, some astonishing and some alarming, but it is still too early to make conclusive decisions. Even within mathematics, absolute truth is problematic and debatable, as showcased by Gödel's incompleteness theorems or the paradoxes of Russell and Skolem, so why should it be any different here? Hence, as educators and policymakers, even as users, let us have a qubit-like state rather than a binary one when it comes to our positions, at least until the AI dust settles. Let us acknowledge that there are currently blessings and curses with the use of AI — be it with LLMs, AI tutors, or adaptive platforms, among other tools — and decisions on what to adopt should be left to our own agency. It is critical to note here that education practitioners and researchers should not remain mere consumers of commercial AI products but should also be involved in the development of suitable AI-related tools and features, tailor-made for education. Overall, it is time to reflect scientifically about these matters and act responsibly, logically, and quickly without rushing, so that in the next few years we reap benefits rather than regret missed opportunities or certain decisions.

What AI brought to the discussion table is way more than technology and LLMs, and this is essential to highlight. AI could be a medium of influence on various levels, and not simply a collection of prompts and tools, just like the Gutenberg press was perceived many centuries back.² The role played by the religious debates at the time is probably played by the political and ideological ones nowadays, but what is certain is that we live in very volatile times with much more at stake than curricula and assessments. Let us remember here that human intelligence is still driving technological advances, and it is for us to decide what happens next, at least in the foreseeable future. So, when it comes to education, let us take carefully considered actions that will genuinely improve it rather than rush into robotising the current systems, fearing that someone will do it before us, and risking doing more harm than good. Failure in this context is not about how many technologies we omit in a classroom; it is how many children we fail to reach and how many learners we leave behind. Eventually, it is not about how fast we include some form of AI in education; it is about how well we do it, if at all. It is not just about how advanced the artificial features and add-ons get; it is about how many teachers and learners we empower and how nurtured our human characteristics and values are. After all, technology’s blooming is meaningless when it is not contributing to the flourishing of humanity. This is not a technology debate, it is about where our human civilisation goes next, and therefore it should not be a competition for the largest market share or proving a point, as there is much more at stake and we are all getting affected by any outcomes. It is not a debate about the tools; it is rather about core values. Let us act accordingly.

Join the conversation: You can tweet us @CambridgeMaths or comment below.