For several months now, research updates and news reports have been circulating about a development in artificial intelligence (AI) that is creating waves: GPT Chat, a robot with which the user can “dialogue”. A striking feature of this chatbot, like other AI developments, is that it seems to know everything: questions across many domains of knowledge are met with detailed, articulate answers. Professionals  and academics around the world representing a variety of different disciplines are questioning how it impacts them and their fields, and specialists are meeting to discuss how to respond to this tool that increasingly affects them.

In the field of music, for example, there is debate over questions such as who owns the rights to a song that is written with the help of an AI application (which bases its output on existing digital information), and how human creativity and its boundaries are to be defined in view of this new reality. Similar issues arise from the repercussions and impacts of GPT Chat specifically, and AI in general, in other areas of culture.

In this context we need to ask ourselves, how does a robot that seems to have the answers to everything, and that it is within everyone's reach, impact the educational field? What happens when students need only to access a chat in order to obtain the answer to any question, or to prepare any assignment?

To begin to familiarize ourselves with this tool and understand its impact in greater depth, we spoke with Dr. Marcelo Dorfsman, Director of the International Master's Degree in Education at the Melton Center, and a specialist in technology-mediated teaching and distance learning.

First of all, how did you first hear of GPT Chat and what this invention is all about?

M.D.: Hebrew University researchers are in constant contact with organizations that constantly send them alerts, invitations and news about various issues related to their research areas. At the beginning of January 2023, I received an invitation from the Mofet Institute, a center for research and teacher training in Israel, to participate in a Webinar on the topic of Chat GPT, which had gone public at the end of 2022, as well as other developments in AI.

I became interested in this topic and explored online sources and bibliography to investigate it further. What is GPT Chat? It literally works like a chat, where a personal written link is established with a robot programmed with diverse and complex logarithms, equipping it to provide information and different types of answers to questions and problems. The native language of GPT Chat is English, although it speaks a very good Spanish. It manages in Hebrew, but doesn’t get the language quite right.

The characteristic of the Chat is that its algorithm allows it to gather information and to learn cumulatively. The more it interacts with people, the more it “learns”, and the more it “knows”. It is important to emphasize that when we talk about a chatbot learning or knowing, we should not think of these actions it in human terms. The robot does not understand, but responds in accordance with an algorithm that is based on the accumulation of data.

Tell us more about what this tool means for us as educators, and what the dangers are.

M.D.: Chat is a tool that allows us to access a lot of information and that can make a lot of things easier for us. For example, we can ask it to put together texts based on certain premises. Since the response will not always be completely correct, we can continue the interaction by clarifying what modifications we want it to make, and thereby improve the result.

First of all, we must understand that we must be careful when using it; what we ask and how we ask are important. Nothing guarantees us that the information we obtain is 100% correct, nor that it is in line with what we want to find out. What do we mean by this? On the one hand, the robot can make mistakes – and often does. On the other hand, it is a group of programmers who make the information available to it, so it will inevitably be biased by the criteria they have used. Although the Chat, through its use, learns and "rearranges" the information, there is still human bias in the initiation of the AI learning process. This requires us to be critical in the use of the information it provides. We cannot treat it as if it were the absolute truth.

What does all this mean for educators?

M.D.: We have an immense responsibility towards our students, to ensure that they are critical of this tool and that they learn to use it correctly. Today there is talk of new skills for artificial intelligence. It is about understanding what we have to know and what we have to transmit to our students, so that they understand what AI is and can use it correctly, thus avoiding a situation in which AI drives us. GPT Chat forces educators up a notch.

It requires more elaboration in what we ask our students, since tasks such as summaries or simple answers can be obtained by asking a question in the Chat. The presence of the GPT Chat invites us to turn our attention to more complex, higher-order tasks such as analysis, contrast, comparison, evaluation. Not simply understanding, explaining, or summarizing. The Chat can be incorporated into the activities so that its use in the learning process is practical. Young people are already using it and it is important to show them that their educators know this tool, that we are also at the forefront and seeking to make an impact on a day-to-day basis.

Understanding that GPT Chat has a significant impact on education, the question arises as to whether it should be included in teacher training.

M.D.: I believe that the teacher training curriculum should be adjusted to incorporate content related to technological training - not only from the perspective of digital literacy, as has been the case up to now, but from the perspective of artificial intelligence, too. The technological has to be a central dimension, along with the disciplinary and the pedagogical.

Following on from the previous question, can you tell us how artificial intelligence is going to be incorporated into the Masters program?

One of the courses that I teach in the program is about educational technologies, and it is there that a module dedicated especially to this subject will be incorporated. A study group is being opened with some of our graduates to research this further and create the curriculum.

This conversation has helped us to understand, first of all, what this tool is about and what we must be attentive to in exploring it. There is still much to investigate. As mentioned, questions arise in relation to the impact of Chat in educational settings and in society in general. How do we "go up a notch" as educators to incorporate this tool? Is there a didactic that we should develop to teach students how to use it? How do we get our students to develop the critical judgment to detect when an answer that has been provided in the chat has a biased perspective? How do we continue working on written production when students have a tool available that can write texts for them? How can we upgrade the assignments that we set and the work that students submit?