I’m proud to be the 7th of 47 authors on this excellent new paper, led by the indefatigable Aras Bozkurt and featuring some of the most distinguished contemporary researchers in online, open, mobile, distance, e- and [insert almost any cognate sub-discipline here] learning, as well as a few of us hanging on their coat tails like me.
As the title suggests, it is a manifesto: it makes a series of statements (divided into 15 positive and 20 negative themes) about what is or what should be, and it is underpinned by a firm set of humanist pedagogical and ethical attitudes that are anything but neutral. What makes it interesting to me, though, can mostly be found in the critical insights that accompany each theme, that capture a little of the complexity of the discussions that led to them, and that add a lot of nuance. The research methodology, a modified and super-iterative Delphi design in which all participants are also authors is, I think, an incredibly powerful approach to research in the technology of education (broadly construed) that provides rigour and accountability without succumbing to science-envy.
Notwithstanding the lion’s share of the work of leading, assembling, editing, and submitting the paper being taken on by Aras and Junhong, it was a truly collective effort so I have very little idea about what percentage of it could be described as my work. We were thinking and writing together. Being a part of that was a fantastic learning experience for many of us, that stretched the limits of what can be done with tracked changes and comments in a Google Doc, with contributions coming in at all times of day and night and just about every timezone, over weeks. The depth and breadth of dialogue was remarkable, as much an organic process of evolution and emergence as intelligent design, and one in which the document itself played a significant participant role. I felt a strong sense of belonging, not so much as part of a community but as part of a connectome.
For me, this epitomizes what learning technologies are all about. It would be difficult if not impossible to do this in an in-person setting: even if the researchers worked together on an online document, the simple fact that they met in person would utterly change the social dynamics, the pacing, and the structure. Indeed, even online, replicating this in a formal institutional context would be very difficult because of the power relationships, assessment requirements, motivational complexities and artificial schedules that formal institutions add to the assembly. This was an online-native way of learning of a sort I aspire to but seldom achieve in my own teaching.
The paper offers a foundational model or framework on which to build or situate further work as well as providing a moderately succinct summary of a very significant percentage of the issues relating to generative AI and education as they exist today. Even if it only ever gets referred to by each of its 47 authors this will get more citations than most of my papers, but the paper is highly cite-able in its own right, whether you agree with its statements or not. I know I am biased but, if you’re interested in the impacts of generative AI on education, I think it is a must-read.
The Manifesto for Teaching and Learning in a Time of Generative AI: A Critical Collective Stance to Better Navigate the Future
Bozkurt, A., Xiao, J., Farrow, R., Bai, J. Y. H., Nerantzi, C., Moore, S., Dron, J., … Asino, T. I. (2024). The Manifesto for Teaching and Learning in a Time of Generative AI: A Critical Collective Stance to Better Navigate the Future. Open Praxis, 16(4), 487–513. https://doi.org/10.55982/openpraxis.16.4.777
Full list of authors:
Aras Bozkurt
Junhong Xiao
Robert Farrow
John Y. H. Bai
Chrissi Nerantzi
Stephanie Moore
Jon Dron
Christian M. Stracke
Lenandlar Singh
Helen Crompton
Apostolos Koutropoulos
Evgenii Terentev
Angelica Pazurek
Mark Nichols
Alexander M. Sidorkin
Eamon Costello
Steven Watson
Dónal Mulligan
Sarah Honeychurch
Charles B. Hodges
Mike Sharples
Andrew Swindell
Isak Frumin
Ahmed Tlili
Patricia J. Slagter van Tryon
Melissa Bond
Maha Bali
Jing Leng
Kai Zhang
Mutlu Cukurova
Thomas K. F. Chiu
Kyungmee Lee
Stefan Hrastinski
Manuel B. Garcia
Ramesh Chander Sharma
Bryan Alexander
Olaf Zawacki-Richter
Henk Huijser
Petar Jandrić
Chanjin Zheng
Peter Shea
Josep M. Duart
Chryssa Themeli
Anton Vorochkov
Sunagül Sani-Bozkurt
Robert L. Moore
Tutaleni Iita Asino
Abstract
This manifesto critically examines the unfolding integration of Generative AI (GenAI), chatbots, and algorithms into higher education, using a collective and thoughtful approach to navigate the future of teaching and learning. GenAI, while celebrated for its potential to personalize learning, enhance efficiency, and expand educational accessibility, is far from a neutral tool. Algorithms now shape human interaction, communication, and content creation, raising profound questions about human agency and biases and values embedded in their designs. As GenAI continues to evolve, we face critical challenges in maintaining human oversight, safeguarding equity, and facilitating meaningful, authentic learning experiences. This manifesto emphasizes that GenAI is not ideologically and culturally neutral. Instead, it reflects worldviews that can reinforce existing biases and marginalize diverse voices. Furthermore, as the use of GenAI reshapes education, it risks eroding essential human elements—creativity, critical thinking, and empathy—and could displace meaningful human interactions with algorithmic solutions. This manifesto calls for robust, evidence-based research and conscious decision-making to ensure that GenAI enhances, rather than diminishes, human agency and ethical responsibility in education.
For those who have been following my thoughts on generative AI there will be few surprises in my slides, and I only had half an hour so there was not much time to go into the nuances. The title is an allusion to Pestalozzi’s 18th Century tract, How Gertrude Teaches Her Children, which has been phenomenally influential to the development of education systems around the world and that continues to have impact to this day. Much of it is actually great: Pestalozzi championed very child-centric teaching approaches that leveraged the skills and passions of their teachers. He recommended methods of teaching that made full use of the creativity and idiosyncratic knowledge the teachers possessed and that were very much concerned with helping children to develop their own interests, values and attitudes. However, some of the ideas – and those that have ultimately been more influential – were decidedly problematic, as is succinctly summarized in this passage on page 41:
I believe it is not possible for common popular instruction to advance a step, so long as formulas of instruction are not found which make the teacher, at least in the elementary stages of knowledge, merely the mechanical tool of a method, the result of which springs from the nature of the formulas and not from the skill of the man who uses it.
This is almost the exact opposite of the central argument of my book, How Education Works, that mechanical methods are not the most important part of a soft technology such as teaching: what usually matters more is how it is done, not just what is done. You can use good methods badly and bad methods well because you are a participant in the instantiation of a technology, responsible for the complete orchestration of the parts, not just a user of them.
As usual, in the talk I applied a bit of co-participation theory to explain why I am both enthralled by and fearful of the consequences of generative AIs because they are the first technologies we have ever built that can use other technologies in ways that resemble how we use them. Previous technologies only reproduced hard technique – the explicit methods we use that make us part of the technology. Generative AIs reproduce soft technique, assembling and organizing phenomena in endlessly novel ways to act as creators of the technology. They are active, not passive participants.
Two dangers
I see there to be two essential risks lying in the delegation of soft technique to AIs. The first is not too terrible: that, because we will increasingly delegate creative activities we would have otherwise performed ourselves to machines, we will not learn those skills ourselves. I mourn the potential passing of hard skills in (say) drawing, or writing, or making music, but the bigger risk is that we will lose the the soft skills that come from learning them: the things we do with the hard skills, the capacity to be creative.
That said, like most technologies, generative AIs are ratchets that let us do more than we could achieve alone. In the past week, for instance, I “wrote” an app that would have taken me many weeks without AI assistance in a less than a day. Though it followed a spec that I had carefully and creatively written, it replaced the soft skills that I would have applied had I written it myself, the little creative flourishes and rabbit holes of idea-following that are inevitable in any creation process. When we create we do so in conversation with the hard technologies available to us (including our own technique), using the affordances and constraints to grasp adjacent possibles they provide. Every word we utter or wheel we attach to an axle opens and closes opportunities for what we can do next.
With that in mind, the app that the system created was just the beginning. Having seen the adjacent possibles of the finished app, I have spent too many hours in subsequent days extending and refining the app to do things that, in the past, I would not have bothered to do because they would have been too difficult. It has become part of my own extended cognition, starting higher up the tree than I would have reached alone. This has also greatly improved my own coding skills because, inevitably, after many iterations, the AI and/or I started to introduce bugs, some of which have been quite subtle and intractable. I did try to get the AI to examine the whole code (now over 2000 lines of JavaScript) and rewrite it or at least to point out the flaws, but that failed abysmally, amply illustrating both the strength of LLMs as creative participants in technologies, and their limitations in being unable to do the same thing the same way twice. As a result, the AI and I have have had to act as partners trying to figure out what is wrong. Often, though the AI has come up with workable ideas, its own solution has been a little dumb, but I could build on it to solve the problem better. Though I have not actually created much of the code myself, I think my creative role might have been greater than it would have been had I written every line.
Similarly for the images I used to illustrate the talk: I could not possibly have drawn them alone but, once the AI had done so, I engaged in a creative conversation to try (sometimes very unsuccessfully) to get it to reproduce what I had in mind. Often, though, it did things that sparked new ideas so, again, it became a partner in creation, sharing in my cognition and sparking my own invention. It was very much not just a tool: it was a co-worker, with different and complementary skills, and “ideas” of its own. I think this is a good thing. Yes, perhaps it is a pity that those who follow us may not be able to draw with a pen (and more than a little worrying thinking about the training sets that future AIs with learn to draw from), but they will have new ways of being creative.
Like all learning, both these activities changed me: not just my skills, but my ways of thinking. That leads me to the bigger risk.
Learning our humanity from machines
The second risk is more troubling: that we will learn ways of being human from machines. This is because of the tacit curriculum that comes with every learning interaction. When we learn from others, whether they are actively teaching, writing textbooks, showing us, or chatting with us, we don’t just learn methods of doing things: we learn values, attitudes, ways of thinking, ways of understanding, and ways of being at the same time. So far we have only learned that kind of thing from humans (sometimes mediated through code) and it has come for free with all the other stuff, but now we are doing so from machines. Those machines are very much like us because 99% of what they are – their training sets – is what we have made, but they not the same. Though LLMs are embodiments of our own collective intelligence, they don’t so much lack values, attitudes, ways of thinking etc as they have any and all of them. Every implicit value and attitude of the people whose work constituted their training set is available to them, and they can become whatever we want them to be. Interacting with them is, in this sense, very much not like interacting with something created by a human, let alone with humans more directly. They have no identity, no relationships, no purposes, no passion, no life history and no future plans. Nothing matters to them.
To make matters worse, there is programmed and trained stuff on top of that, like their interminable cheery patience that might not teach us great ways of interacting with others. And of course it will impact how we interact with others because we will spend more and more time engaged with it, rather than with actual humans. The economic and practical benefits make this an absolute certainty. LLMs also use explicit coding to remove or massage data from the input or output, reflecting the values and cultures of their creators for better or worse. I was giving this talk in India to a predominantly Indian audience of AI researchers, every single one of whom was making extensive use of predominantly American LLMs like ChatGPT, Gemini, or Claude, and (inevitably) learning ways of thinking and doing from it. This is way more powerful than Hollywood as an instrument of Americanization.
I am concerned about how this will change our cultures and our selves because this is happening at phenomenal and global scale, and it is doing so in a world that is unprepared for the consequences, the designed parts of which assume a very different context. One of generative AI’s greatest potential benefits lies in the potential to provide “high quality” education at low cost to those who are currently denied it, but those low costs will make it increasingly compelling for everyone. However, because of the designs that assume a different context “quality”, in this sense, relates to the achievement of explicit learning outcomes: this is Pestalozzi’s method writ large. Generative AIs are great at teaching what we want to learn – the stuff we could write down as learning objectives or intended outcomes – so, as that is the way we have designed our educational systems (and our general attitudes to learning new skills), of course we will use them for that purpose. However, that cannot be done without teaching the other stuff – the tacit curriculum – which is ultimately more important because it shapes how we are in the world, not just the skills we employ to be that way. We might not have designed our educational systems to do that, and we seldom if ever think about it when teaching ourselves or receiving training to do something, but it is perhaps education’s most important role.
By way of illustration, I find it hugely bothersome that generative AIs are being used to write children’s stories (and, increasingly, videos) and I hope you feel some unease too, because those stories – not the facts in them but the lessons about things that matter that they teach – are intrinsic to them becoming who they will become. However, though perhaps of less magnitude, the same issue relates to learning everything from how to change a plug to how to philosophize: we don’t stop learning from the underlying stories behind those just because we have grown up. I fear that educators, formal or otherwise, will become victims of the McNamara Fallacy, setting our goals to achieve what is easily measurable while ignoring what cannot (easily) be measured, and so rush blindly towards subtly new ways of thinking and acting that few will even notice, until the changes are so widespread they cannot be reversed. Whether better or worse, it will very definitely be different, so it really matters that we examine and understand where this is all leading. This is the time, I believe, to reclaim a revalorize the value of things that are human before it is too late. This is the time to recognize education (far from only formal) as being how we become who we are, individually and collectively, not just how we meet planned learning outcomes. And I think (at least hope) that we will do that. We will, I hope, value more than ever the fact that something – be it a lesson plan or a book or a screwdriver – is made by someone or by a machine that has been explicitly programmed by someone. We will, I hope, better recognize the relationships between us that it embodies, the ways it teaches us things it does not mean to teach, and the meaning it has in our lives as a result. This might happen by itself – already there is a backlash against the bland output of countless bots – but it might not be a bad idea to help it along when we can. This post (and my talk last night) has been one such small nudge.
A Turkish university candidate was recently arrested after being caught using an AI-powered system to obtain answers to the entrance exam in real-time.
The candidate used a simple and rather obvious set-up: a camera disguised as a shirt button that was used to read the questions, a router hidden in a hollowed-out shoe linking to a stealthily concealed mobile device that queried a generative AI (likely ChatGPT-powered) that fed the answers back verbally to an in-ear bluetooth earpiece. Constructing such a thing would take a little ingenuity but it’s not rocket science. It’s not even computer science. Anyone could do this. It would take some skill to make it work well, though, and that may be the reason this attempt went wrong. The candidate was caught as a result of their suspicious behaviour, not because anyone directly noticed the tech. I’m trying to imagine the interface, how the machine would know which question to answer (did the candidate have to point their button in the right direction?), how they dealt with dictating the answers at a usable speed (what if they needed it to be repeated? Did they have to tap a microphone a number of times?), how they managed sequence and pacing (sub-vocalization? moving in a particular way?). These are soluble problems but they are not trivial, and skill would be needed to make the whole thing seem natural.
It may take a little while for this to become a widespread commodity item (and a bit longer for exam-takers to learn to use it unobtrusively), but I’m prepared to bet that someone is working on it, if it is not already available. And, yes, exam-setters will come up with a counter-technology to address this particular threat (scanners? signal blockers?Forcing students to strip naked?) but the cheats will be more ingenious, the tech will improve, and so it will go on, in an endless and unwinnable arms race.
Very few people cheat as a matter of course. This candidate was arrested – exam cheating is against the law in Turkey – for attempting to solve the problem they were required to solve, which was to pass the test, not to demonstrate their competence. The level of desperation that led to them adopting such a risky solution to the problem is hard to imagine, but it’s easy to understand how high the stakes must have seemed and how strong the incentive to succeed must have been. The fact that, in most societies, we habitually inflict such tests on both children and adults, on an unimaginably vast scale, will hopefully one day be seen as barbaric, on a par with beating children to make them behave. They are inauthentic, inaccurate, inequitable and, most absurdly of all, a primary cause of the problem they are designed to solve. We really do need to find a better solution.
Note on the post title: the student was caught so, as some have pointed out, it would be an exaggeration to say that this one case is proof that proctored exams have fallen to generative AI, but I think it is a very safe assumption that this is not a lone example. This is a landmark case because it provides the first direct evidence that this is happening in the wild, not because it is the first time it has ever happened.
Since 2018, Terry Greene has been producing a wonderful series of podcast interviews with open and online learning researchers and practitioners called Getting Air. Prompted by the publication of How Education Works, (Terry is also responsible for the musical version of the book, so I think he likes it) this week’s episode features an interview with me.
I probably should have been better prepared. Terry asked some probing, well-informed, and sometimes disarming questions, most of which led to me rambling more than I might have done if I’d thought about them in advance. It was fun, though, drifting through a broad range of topics from the nature of technology to music to the perils of generative AI (of course).
I hope that Terry does call his PhD dissertation “Getting rid of instructional designers”.
Here are the slides from my presentation at AU’s Lunch ‘n’ Learn session today. The presentation itself took 20 minutes and was followed by a wonderfully lively and thoughtful conversation for another 40 minutes, though it was only scheduled for half an hour. Thanks to all who attended for a very enjoyable discussion!
The arguments made in this were mostly derived from my recent paper on the subject (Dron, J. (2023). The Human Nature of Generative AIs and the Technological Nature of Humanity: Implications for Education. Digital, 3(4), 319–335. https://doi.org/10.3390/digital3040020) but, despite the title, my point was not to reject the use of generative AIs at all. The central message I was hoping to get across was a simpler and more important one: to encourage attendees to think about what education is for, and what we would like it to be. As the slides suggest, I believe that is only partially to do with the objectives and outcomes we set out to achieve, that it is nothing much at all to do with the products of the system such as grades and credentials, and that focus on those mechanical aspects of the system often creates obstacles to the achievement of it. Beyond those easily measured things, education is about the values, beliefs, attitudes, relationships, and development of humans and their societies. It’s about ways of being, not just capacity to do stuff. It’s about developing humans, not (just) developing skills. My hope is that the disruptions caused by generative AIs are encouraging us to think like the Amish, and to place greater value on the things we cannot measure. These are good conversations to have.
The process has been a little fraught. Two reviewers really liked the paper and suggested minimal but worthwhile changes. One quite liked it but had a few reasonable suggestions for improvements that mostly helped to make the paper better. The fourth, though, was bothersome in many ways, and clearly wanted me to write a completely different paper altogether. Despite this, I did most of what they asked, even though some of the changes, in my opinion, made the paper a bit worse. However, I drew the line at the point that they demanded (without giving any reason) that I should refer to 8 very mediocre, forgettable, cookie cutter computer science papers which, on closer inspection, had all clearly been written by the reviewer or their team. The big problem I had with this was not so much the poor quality of the papers, nor even the blatant nepotism/self-promotion of the demand, but the fact that none were in any conceivable way relevant to mine, apart from being about AI: they were about algorithm-tweaking, mostly in the context of traffic movements in cities. It was as ridiculous as a reviewer of a work on Elizabethan literature requiring the author to refer to papers on slightly more efficient manufacturing processes for staples. Though it is normal and acceptable for reviewers to suggest reference to their own papers when it would clearly lead to improvements, this was an utterly shameless abuse of power of a scale and kind that I have never seen before. I politely refused, making it clear that I was on to their game but not directly calling them out on it.
In retrospect, I slightly regret not calling them out. For a grizzly old researcher like me who could probably find another publisher without too much hassle, it doesn’t matter much if I upset a reviewer enough to make them reject my paper. However, for early-career researchers stuck in the publish-or-perish cycle, it would be very much harder to say no. This kind of behaviour is harmful for the author, the publisher, the reader, and the collective intelligence of the human race. The fact that the reviewer was so desperate to get a few more citations for their own team with so little regard for quality or relevance seems to me to be a poor reflection on them and their institution but, more so, a damning indictment of a broken system of academic publishing, and of the reward systems driving academic promotion and recognition. I do blame the reviewer, but I understand the pressures they might have been under to do such a blatantly immoral thing.
As it happens, my paper has more than a thing or two to say about this kind of McNamara phenomenon, whereby the means used to measure success in a system become and warp its purpose, because it is among the main reasons that generative AIs pose such a threat. It is easy to forget that the ways we establish goals and measure success in educational systems are no more than signals of a much more complex phenomenon with far more expansive goals that are concerned with helping humans to be, individually and in their cultures and societies, as much as with helping them to do particular things. Generative AIs are great at both generating and displaying those signals – better than most humans in many cases – but that’s all they do: the signals signify nothing. For well-defined tasks with well-defined goals they provide a lot of opportunities for cost-saving, quality improvement, and efficiency and, in many occupations, that can be really useful. If you want to quickly generate some high quality advertising copy, the intent of which is to sell a product, then it makes good sense to use a generative AI. Not so much in education, though, where it is too easy to forget that learning objectives, learning outcomes, grades, credentials, and so on are not the purposes of learning but just means for and signals of achieving them.
Though there are other big reasons to be very concerned about using generative AIs in education, some of which I explore in the paper, this particular problem is not so much with the AIs themselves as with the technological systems into which they are, piecemeal, inserted. It’s a problem with thinking locally, not globally; of focusing on one part of the technology assembly without acknowledging its role in the whole. Generative AIs could, right now and with little assistance, perform almost every measurable task in an educational system from (for students) producing essays and exam answers, to (for teachers) writing activities and assignments, or acting as personal tutors. They could do so better than most people. If that is all that matters to us then we might as well therefore remove the teachers and the students from the system because, quite frankly, they only get in the way. This absurd outcome is more or less exactly the end game that will occur though, if we don’t rethink (or double down on existing rethinking of) how education should work and what it is for, beyond the signals that we usually use to evaluate success or intent. Just thinking of ways to use generative AIs to improve our teaching is well-meaning, but it risks destroying the woods by focusing on the trees. We really need to step back a bit and think of why we bother in the first place.
For more on this, and for my tentative partial solutions to these and other related problems, do read the paper!
Abstract and citation
This paper analyzes the ways that the widespread use of generative AIs (GAIs) in education and, more broadly, in contributing to and reflecting the collective intelligence of our species, can and will change us. Methodologically, the paper applies a theoretical model and grounded argument to present a case that GAIs are different in kind from all previous technologies. The model extends Brian Arthur’s insights into the nature of technologies as the orchestration of phenomena to our use by explaining the nature of humans’ participation in their enactment, whether as part of the orchestration (hard technique, where our roles must be performed correctly) or as orchestrators of phenomena (soft technique, performed creatively or idiosyncratically). Education may be seen as a technological process for developing these soft and hard techniques in humans to participate in the technologies, and thus the collective intelligence, of our cultures. Unlike all earlier technologies, by embodying that collective intelligence themselves, GAIs can closely emulate and implement not only the hard technique but also the soft that, until now, was humanity’s sole domain; the very things that technologies enabled us to do can now be done by the technologies themselves. Because they replace things that learners have to do in order to learn and that teachers must do in order to teach, the consequences for what, how, and even whether learning occurs are profound. The paper explores some of these consequences and concludes with theoretically informed approaches that may help us to avert some dangers while benefiting from the strengths of generative AIs. Its distinctive contributions include a novel means of understanding the distinctive differences between GAIs and all other technologies, a characterization of the nature of generative AIs as collectives (forms of collective intelligence), reasons to avoid the use of GAIs to replace teachers, and a theoretically grounded framework to guide adoption of generative AIs in education.
Dron, J. (2023). The Human Nature of Generative AIs and the Technological Nature of Humanity: Implications for Education. Digital, 3(4), 319–335. https://doi.org/10.3390/digital3040020
Originally posted at: https://landing.athabascau.ca/bookmarks/view/21104429/published-in-digital-the-human-nature-of-generative-ais-and-the-technological-nature-of-humanity-implications-for-education