BECOME A MEMBER! Sign up for TIE services now and start your international school career


Artificial Intelligence and the Automaticity of Skill Development

By Conrad Hughes
Artificial Intelligence and the Automaticity of Skill Development

When the ChatGPT news started to go viral, the immediate reaction in education circles, in essence, was to say (again!) that assessments that can be completed by an algorithm probably aren't fit for purpose. As such, Victorian-styled tasks built on regurgitating memory should be replaced by those requiring more distinctly human traits of intelligence (interpersonal and intrapersonal intelligence, criticality, and creativity).

Conservative approaches to learning (wooden homework tasks, the reinforcement of routine lower order cognitive tasks, and heavy testing) are playing to constructs that can, effectively, be outsourced by machine learning. By now, most of us have seen something written by ChatGPT and it is clear that if the goal is for humans to write at that level, there is a problem, and we might as well give up on trying to get students to achieve the task.

This would be ridiculous of course, since it is not the mental product that is the point of these exercises in education, it is the process behind them. The products are merely the outcome, not the journey, and it's not because the outcome can be achieved some other way or entity that it is useless. 

In other words, this does not mean that we don't need to build up the intellectual muscle that can be duplicated by a machine; common sense and the most basic rules of cognition remind us that reinforcing long-term memory, building automaticity in numeracy and literacy, and old-fashioned learning off by heart are unavoidable if you want to be good at something. You won't become a great athlete, musician, or public speaker without practice. Vulgarized and oversimplified studies on deliberate practice nonetheless point to the truism that without hard work, of a fairly low level machine-learning type, little will be achieved. Deliberate practice in particular means that it is not just trudging through hours of repetition that causes improvement, but analyzing weaknesses and focusing on precisely those elements, to practice them into near perfection. This is very much the premise of so-called "deep learning," a process whereby machine learning will steadily improve performance as the data set of examples increases and the patterning of what the best solution or approach is becomes clearer. This type of adaptive, recursive behavior is really at the root of what intelligence in its most generic form represents, improvement through analysis.

So, it's not because a machine can do it that we should stop doing it altogether; there are many aspects of human behavior that have been, are, and will be outsourceable, but they are still quintessential and cannot be disbanded. If a car can drive faster than a human can run, should we stop running? If you can find the same amount of nutrients, vitamins, and minerals through a machine-regulated drip, should you stop eating? If samples can reproduce drum beats as well as humans, should there be no more drummers? 

So yes, assessments should be reformed, and we should continue to look for more inspiring, human types of intelligence to celebrate, but ChatGPT and other powerful types of machine learning should not stop us from wanting to improve ourselves, albeit often and necessarily through measures that are actually very mechanical and could be performed more efficiently by an algorithm. Ultimately, the fact that artificial intelligence can complete the same tasks human beings can should not be seen as a threat. We can deconstruct the scenario completely, reverse it, and set a new premise: what do I as a human being need to do as well as, and perhaps even better than a machine, to improve myself? Human beings are not robots but do seek some form of self-actualization that requires constructs such as practice, reinforcement consistency, and adaptive learning. Let’s embrace that.

Read more about education technology in Artificial Intelligence and Society: Implications for Educators.

Conrad Hughes (MA, PhD, EdD) is the Director General of the International School of Geneva. He has been a school principal, director of education, International Baccalaureate diploma program coordinator, and teacher in schools in Switzerland, France, India, and the Netherlands. Conrad is a Senior Fellow at UNESCO's International Bureau of Education, a member of the advisory board for the University of the People, and a research assistant at the University of Geneva's department of psychology and education. He teaches philosophy.


Please fill out the form below if you would like to post a comment on this article:


04/29/2023 - Baela
How do you reconcile the layers of academic and non-academic learning in routine learning tasks with AI machine learning, as a researcher of schools2030 these are critical questions for holistic approaches to learning acquistion skills