What's Still Wrong With Education
One Size Fits All
In "What's Wrong With Education", a blog from February 2022, we argued that the single greatest deficiency in education is the lack of a focus on the driving force afforded by the personal interests of every child. Because the nature of education until now has necessitated grouping children and young adults together in relatively large groups in order to manage the scale of the problem - not enough teachers and many learners - we have been forced to teach groups together, move them through education at roughly the same speed, assess them together and comparatively, and so produce a system that is to a very large extent "one size fits all". That there are different kinds of school, different examination specifications and that there is gradual diversification mitigates this a little, but at the end of every educational process there sits an assessment process that is monolithic and exercises a constraining influence - a "dead hand" - over everything that leads towards it.
There are other mitigating factors such as the diversity of the students themselves, the infinite variation that occurs in classes and between different groups of students and their teachers, and the additional variation that inevitably arises from cultural and geographical differences. All this should be acknowledged as having supplied some elements of variation despite the monolithic nature of the curriculum and of assessment. But that curriculum and assessment remains as a constraining factor on learning that alienates an enormous proportion of young people from education.
The Human Cost
The human cost of this process has been incalculable. It has not simply been a matter of children being forced to learn what they saw no need to learn at inappropriate speeds in ways unsuited to their learning-requirements and styles. Far more significant even than that is the way the process for a multiplicity of reasons has alienated huge proportions of every generation from learning itself. One consequence of that has been that the project of encouraging "lifelong learning" has had to fight an uphill battle, and that for every success there have been many more failures.
One reason why this system has persisted for centuries despite its many disadvantages is that we have largely endorsed a view that says that "your elders know best", that it is not only legitimate but desirable and necessary for the content, methods, speed and direction of educational travel to be determined by students' "elders and betters". No doubt the absence of a viable alternative allowed that assumption to go unchallenged, but as we have suggested in "What Price Education?", things are about to change and, to some extent, have already changed.
Personal Educational Trajectories
Leaving aside for a moment the numerous objections to a personalised educational trajectory that will be voiced by the educational establishment - those, that is, with a vested interest in things remaining essentially the same - we may reasonably suppose that technology like chatGPT at the very least increases the possibility and likelihood of personalised education becoming tractable.
Very young children will almost certainly still need and benefit from education in nurseries and primary schools, but from the age when they become better-able to exercise autonomy, more clear about their interests and the direction they want their lives to take, the advantages to be had from giving them more control over their own education are numerous and significant.
Somewhere between the ages of six and eight - there is no fixed age because different children will develop at different speeds in different ways, itself as we have said one of the problems with the education we already have - children start to exhibit preferences for what they learn, how they learn, when they learn, at what speed they learn, and a host of other educational parameters that should be taken into account but usually are not.
The global name we are giving this is "Interest-Driven Education" because the key factor in it is that we grant children more and more say in what they learn, how they learn, when they learn, at what speed they learn, and so forth. We granted them "ownership" of their own learning.
In a traditional classroom this individualised learning is almost impossible to realise, and the larger the class, the greater the difficulty. The most enlightened teachers manage some kind of personalised learning even in the most difficult conditions where classes are upwards of 30, 40, even as many as 80 children, and the term we have used for this in our educational theory is "differentiation". But differentiation can do only so much when children are grouped by age, all expected to be in one place, all expected to listen to one voice, and so forth. The idea that each could be allowed to pursue their own interests and to determine - or at least have a significant say in - all the educational parameters we have mentioned seems little short of absurd in such conditions, and so it almost certainly is.
ChatGPT and its Successors
As we have indicated in "What Price Education?", an AI engine like OpenAI's chatGPT could, if provided with a capacity to "remember", by which we really mean "record and log" the learning of each student or user, could without undue difficulty then create a "curriculum map" that indicated what a child's learning had covered and what it had not. By setting personalised assessments based on these learning-logs each child's grasp of the material it had putatively learnt could be measured, which would facilitate better learning and indicate areas of strength and weakness.
The role of the teacher will change rather than disappear in such a process because navigating the maze of information and the multiple levels of difficulty and sophistication that a chatbot makes available will not be something that any and every child can manage autonomously. Teachers may need to encourage deeper and more persistent inquiry as well as steer children away from things that may be discouraging because they are too difficult. But we also envisage that the process will to a considerable extent be self-correcting and self-limiting: children will migrate away from things when they become too difficult or insufficiently interesting.
The human dimension will not be lost, but the mechanism of content-delivery will fundamentally change because each child - we are back again to learning by self-play that is always perfectly matched to the appropriate standard as with AlphaZero and David Silver in our earlier blog "What Price Education?" - will essentially take responsibility for its own questioning, level of difficulty, direction, scope, depth, speed and navigation.
The principal point here is that no child need be constrained by an externally-imposed sequence of learning, or by a particular level of difficulty, breadth or depth: by following its interests, a child will gradually map a substantial portion of the knowledge-space appropriate to its age without anyone imposing any limits on that space, forcing it to go too far, or preventing it from going further than a traditional curriculum might suppose it was capable of going.
Traditional education suggests that children need to learn in a certain sequence in a certain way at a certain speed and to a certain standard, but chatGPT makes all that unnecessary. All a student need do is ask questions, follow up on the answers provided, and only stop when either its curiosity, interest or the scope of the chatbot is exhausted. Nobody need worry whether the child is "learning enough" because there will no longer be an externally-dictated notion of "enough"; each child will decide for itself in conjunction with the system and a teacher.
The ridiculous, damaging practice of requiring all children to sit the same test at the same age and then assessing their potential on the basis of their performance can be dispatched forever. Children need no longer carry the stigma associated with competitive comparative assessment because assessment will only ever be relative to what targets they have set themselves, and unlike the human administration of education, the results will be informative but not pejorative or, for that matter, adulatory.
Condemnation and Adulation
Conventional, traditional education has a lot to answer for in its attitude to children who "struggle" and others whom it deems "clever". Children are not all equal, have different interests, can do more or less of this and that, and generally differ in so many ways that it is inhuman even to try to compare them. Yet we have based much of our educational provision on motivations afforded by "carrot and stick" and by "praise and condemnation". We pay lip service to the claim that children who "do well" may just have inherited the genes and the environment that have contributed to their "success" while others have been disadvantaged when measured on such scales, but in practice we have reverted to praise and blame, condemnation and adulation whenever we have come to evaluate each student and child.
Assessment based on self-set objectives and standards holds out hope that we might create a system that recognises and grows each child using the results of the education they have designed for themselves. In such an educational paradise every child will really matter, and every child will at least have a chance of being seen to have done well in relation to is natural ability.
Navigating the Knowledge-Space
Educators will claim that by allowing children to decide the direction (etc.) of their learning there will arise enormous gaps in their knowledge because the curriculum of a traditional education will not be covered and so they will not be enabled to take up a profitable place in society, earn a living, or make a useful contribution to an economy. Yet since nobody has ever tried to run education as we are suggesting, this must remain a speculative claim at best. Moreover, all education leaves gaps, and most of us find that we need to go back and learn new things long after we have left school. There is nobody who knows everything about anything, so this riposte by educators needs to be viewed with some scepticism.
Children granted more autonomy over their learning will surprise us with the ingenuity of their questions and the territory they explore. That territory will not necessarily be a part of conventional educational curricular provision, but so what? Isn't one of the other lessons of chatGPT, to which we will return in another blog, that humans need to diversify, to explore new ways of thinking and analysing the world, rather than spending most of their lives learning things that new AI engines will know better and more comprehensively?
A Golden Age
We have the opportunity to embrace a golden age of human learning. Traditional, conventional education will almost certainly try either to suppress or to control that process. We should resist those attempts.