Using Generative Artificial Intelligence in Education 

I’ve recently been in several think tanks and conferences about education and, perhaps not surprisingly, it seems to be basically impossible to talk about school without mentioning ChatGPT. The excitement, awe, and fear about how this will change everything is palpable as it leapfrogs more basic questions that lie largely unanswered. The Covid19 pandemic has made us think of these large-scale changes with a sense of urgency.

What’s the effectiveness of technological aids in learning? How can we use hybrid and fully remote learning systems more effectively? How to use the power of technology to increase access to learning? These and many other such substantial questions are inviting attention and changing the face of global education at various levels.  

There is no doubt that Large Language Models are influencing student learning, and very quickly, so schools need to adapt whether they like it or not. 

Having read the outstanding guidance set out by UNESCO, three essential points appear quite quickly: 

We need to remember that generative artificial intelligence has not been screened or quality-controlled the way that textbooks have. Using it in the classroom means that we are in the hands of automated text that might very easily propagate and exacerbate bias, including racial and gender bias. 

Large Language Models contain mistakes: they are very far from being perfectly reliable and need to be understood this way. 

However, using generative artificial intelligence creatively allows for extended critical thinking, mainly by using prompts, trialing essay titles and examination-styled questions with it, and then critiquing the responses. 

Therefore, it is the prime responsibility of administrators, teachers, and policymakers to err on the side of caution and not rely entirely on automated systems. Human intervention will still be essential and will still dictate how automation is incorporated into education. 

Other points could be added, but I would place those three at the forefront, and in that order: ethics, reliability, and creativity.  

The fundamental implication of this is that teachers should open metacognitive discussions about generative artificial intelligence, particularly Large Language Models like ChatGPT so that there is an intergenerational, critical discourse on what it is and how we might use it. In fact, having contemporaneous discussions of this sort, not only on technology but on other issues such as sustainable development, world peace, and inclusion, creates a healthy environment of sharing, thinking, and co-creating possibilities.