TESTIMONY. “ChatGPT has found its audience, we are on something similar to the arrival of Google in 1998”


ChatGPT is this artificial intelligence that has the answer to almost everything. It can also compile several results from search engines like Google. For the past few weeks, the website where it is available has been receiving a lot of visitors. ChatGPT captivates the public, but it also begins to worry.

Have you heard of ChatGPT in recent weeks? It is an artificial intelligence, accessible on this website. She is able to summarize the results of a search engine, or to answer questions from Internet users on almost any subject.

Our facetious colleagues from France 3 Alsace, for example, asked this “conversational intelligence“if Alsace were to leave the Grand Est region. Thorny question on which ChatGPT played caution, not taking a position. Instead, a text that summarizes the arguments for and against.

Joking aside, the rapid rise in power of this artificial intelligence inevitably raises questions. It fascinates as much as it forces us to adapt, according to specialists in the world of information.

It is a major technological advance, we are on something similar to the arrival of Google in 1998. ChatGPT found its audience in a very short time: in a few weeks, of the order of a million users“, notes Martial Martin. For the teacher-researcher in communication and information sciences at the University of Reims (Marne), the growing place of this tool is interesting to observe.

We must monitor whether ChatGPT, or other conversational intelligences such as those of connected speakers such as Amazon Echo or Google Nest, come to replace traditional search engines. What we observe for ChatGPT is that it coordinates and synthesizes the elements of the results obtained in a Google search for example“, notes the one who is also the director of the University Institute of Technology in Troyes (Aube).

According to the academic, this “conversational intelligence“offers a rather elaborate reasoning.”This can be a facility for a college or high school student who would have difficulties for his research. The reflection is truly arranged on a subject. The agent provides about ten well-argued and well-structured lines”explains Martial Martin.

Caution all the same, on the idea of ​​completely trusting ChatGPT. The researcher qualifies his statement, indicating that he is not immune to manipulation. What relates, for example, this blog specialized in the world of information. “Choices are made in terms of coding. Artificial intelligence is based on a principle of coding. For this coding, there are choices that are made. This can lead to the development of ideologically biased responses, so that this tool can turn into a tool of indoctrination, which passes for transparent“, details the university official.

He calls on the public to be vigilant, and to always check their sources. “The risk is to recognize this intelligence as superior to what it is. A number of users will limit themselves to just that. We must put ChatGPT into context: it is only one element in a larger research. Research must be linked with other sources of information“, he points.

This is also noted by Yann Verchier, at the University of Technology of Troyes (Aube). For this person in charge of the center of educational innovation in this faculty, “even if this program provides relevant answers, the sources may not always be verified, thus causing errors“. Two other reservations are noted by this holder of the Smart Campus/Smart Cursus research chair: “The knowledge of intelligence stops for the moment at the year 2021. Nothing also excludes that it does not learn from false contents and questionable resources on its own“.

Nothing also excludes that she does not learn from herself false contents and questionable resources

Yann Verchier

teacher at the University of Technology of Troyes

The development of this “conversational intelligence“questions a lot, on the professional side.”This will question the practices as a teacher. We will have to quickly rethink the way of working with learners. Quickly explain what interests, what limits and what complements there may be to OpenGPT. We must also determine the professional framework, we can use it in an ethical way“, says Martial Martin. Same concern on the side of the University of Technology of Troyes (Aube). In this establishment, the staff and the students must precisely be made aware soon of good practices for ChatGPT.

Evaluation methods will be questioned, because students have access to a resource that gathers information in real time. On the teachers’ side, the question of relevance arises for all current assessments, whether on paper, digitally or remotely. We demonstrated to them, by doing a test with elements taken from their evaluations inserted into ChatGPT, that a novice who had not taken their courses could get by with this tool. We are really in front of different resources, with a mix and unique responses“, underlines Yann Verchier.

The debate continues for the time being, especially in education circles, to determine the future place of ChatGPT.