- 1 “By making ChatGPT available to all on the 30 November 2022, OpenAI ushered in an area of unbelieva (...)
1More than ever, the rapid growth of generative artificial intelligence1 addresses the issue of the role of social representations. Indeed, if these representations constitute a shared system enabling individuals to interpret and apprehend their personal and professional environment and guide them in their choices, they also question the way in which current educational programs integrate these emerging technologies.
2More particularly, as generative AI plays an increasingly important role in practices (Gillespie et al., 2025), it is rolled out in numerous professional domains (Blons, 2023): “The most affected roles are legal professions (45.2% of jobs potentially impacted), jobs related to information technology and mathematics (44.8%), as well as roles related to business administration and finance (44.6%).”2 Education and training are not spared from these major changes.3 It becomes essential to understand how different communities of learners consider generative AI tools to mobilize them in pedagogical practices.
3Hence, this article raises the following issue: what social representations of generative AI do learners develop according to their education level, and how can these representations be taken into consideration by teachers to support their career choices? To provide an answer, we first present the theoretical framework of social representations, then the implemented exploratory methodology. Following the presentation of the enquiry’s results, a discussion will propose pedagogical approaches to provide informed, critical and inclusive support for professional orientation.
4Social representations (Moscovici, 1976) form a shared system of environmental interpretation, regulation of judgments, behaviors, justifications of definition, and the preservation of this group’s identity (Abirc, 1994). They also cover the “overall of beliefs, knowledge, and opinions that are produced and shared by all individuals of the same group regarding a given social object” (Guimelli, 2009, p. 63). Flament and Rouquette (2003) outline two minimal conditions without which it is impossible to discuss objects of representation: socio-cognitive salience and the existence of object-related practices. Socio-cognitive salience dictates that an object needs to be sufficiently present in a social and cognitive space to be shared, discussed, and recognized by a collective. On this point, we may note that the technologies based on generative AI are regularly addressed in the media, everyday conversations, at school or work. They relate to a generic category comprising a certain number of objects (material, software, prompt-engineered results). Generative AI appears to be a salient object in the sense that everybody has heard something about it and can express an opinion. The existence of object-related practices is to be linked with concrete activities engaging individuals or groups. Concerning generative AI, many of them are used in an educational context: write a text, improve a picture or even create a video. They are embedded in a strongly held sociocultural context. Indeed, the proposed generative AI tools and their types of usage are related to the sociodemographic characteristics (Méndez-Suárez et al., 2023) of individuals, in particular their gender (González-Anleo et al., 2004), their age, their education level (Kacperski et al., 2023). These usages form part of a set of significations and social interactions that inscribe individuals in a life trajectory. Generative AI tools are transforming jobs (Bettache & Foisy, 2019). Everybody must contemplate throughout their career becoming their own entrepreneur, that is to say “create” or “recreate” their post and their activities, independently from the support structure. Hence, they find themselves “overexposed and weakened because they are not supported anymore by collective regulatory systems” (Castel, 2003, p. 44). In this context, consideration must be given to the representations of AI, its potential for causing fear in professional activities among the new generation, through two questions:
-
What social representations do today’s learners have of AI and is there any difference according to their support structure (middle school, high school and university)?
-
How can teachers tackle these representations to support learners’ career choices?
5To answer these questions, we draw upon studies led within the “Cordée de la réussite”4 program. Supported by the Ministry of National Education and the Ministry of Higher Education, Research and Innovation, this scheme consists of a set of front-line actors in different “roped up” [encordée] institutions (from middle school to university) who, through their exchanges and actions, facilitate a collective process of reflection on support and equal opportunities: it provides support (individual and collective) for each learner to undertake higher education or complete a chosen professional integration. While in some discourses, generative AI claims to foster greater inclusion,5 the “Cordée de la réussite” is an appropriate framework for implementing a reflective approach on its role in the professional trajectory of each learner.
- 6 SOAN stands for S’Orienter et s’Approprier le savoir avec le Numérique (orientation and knowledge a (...)
6The implemented methodology is explanatory (Étienne, 2022). It concerns three groups included in the SOAN6 cordée de la réussite:
71. Technology University Bachelor students [Bachelor universitaire de technologie (BUT)] at the Institut Universitaire de Technologie (IUT) of Béziers (France)
- 7 Translator’s note: MMI refers to Métiers du multimédia et de l’Internet (careers in multimedia and (...)
8In the national program (NP, produced and enacted in 2021) related to training (BUT-MMI),7 no information on the inclusion of AI is given. Hence, even if the program of the 24 BUT specializations is supposed to facilitate professional integration, when the NP was written (2021), AI had not been part of the representations of the professionals, nor of the teachers in charge of the elaboration of these NP. The respondents in this study are 60 students from the IUT of Béziers, enrolled in their first year of BUT-MMI, attending a class entitled “Ergonomics and Accessibility.” They were asked to respond to an online questionnaire on the Moodle platform.
92. Middle school students from the collège André-Chanson in Meyrueis (France)
- 8 Bulletin Officiel n° 9, 29 February 2024, ministère de l’Éducation nationale et de la Jeunesse (Min (...)
- 9 “The ministry of National Education announced on Thursday 6 February that middle and high school st (...)
10The 2024 technology program in middle schools8 incorporates AI-related aspects.9 This program specifies that it is important that learners receive “an education that prepares students to face the technological challenges related to social and ecological transition issues.” The same questionnaire was sent to 90 middle school students (between the ages of 11 and 15 years) which they could answer on a voluntary basis.
113. High school students from the lycée Chaptal in Mende (France)
- 10 Bulletin Officiel Annexe: Programme de sciences numériques et technologie de seconde générale et te (...)
12In the Digital Sciences and Technology program for first-year high school students, the use of AI is requested for “the real-time analysis of the environment with various sensory devices (cameras, radars, lidars, etc.).”10 In this context, AI is seen as a technical tool at the service of the implementation and interpretation of physical measuring devices. The questionnaire was sent to 60 high school students (between the ages of 15 to 17 years) which they could answer on a voluntary basis.
13The questionnaire dealt with (Boulan 2015) three major dimensions of generative AI:
-
The relationship between generative AI and everyday elements: association of generative AI with a painting, a color, a film, a book, a music, an individual, a videogame, and gender.
-
Definition and perception of generative AI in different contexts: identification of keywords associated with generative AI and evaluation of its future impact on personal life, work, the environment, and society (using a Likert scale).
-
Usage and utility of generative AI: personal experience with AI (use, frequency and context) and perceived utility of generative AI (Davis, 1989).
14The respondents to the questionnaire are divided as follows:
Table 1
|
Number of participants
|
|
Middle school students
|
14 (including 18.75% of female students)
|
|
High school students
|
48 (including 9.6% of female students)
|
|
University students
|
47 (including 56.81% of female students)
|
Breakdown of respondents.
15This repartition shows a notable imbalance according to the education level. Among middle school students, only 18.75% are teenage girls. This imbalance is even more pronounced for high school students where female pupils represent only 9.6% of the cohort. However, the trend is reversed among university students, with 56.81% of female students, thereby offering a more representative subsample. This may be explained by the questionnaire’s availability for middle and high school students, as opposed to university students whose response to these questions was integrated into the pedagogical approach of a course.
16The different responses to the questionnaire are organized according to the three major dimensions of generative AI as established in the methodology.
17The responses to the question “How would you picture AI if it had to be associated with: a gender (feminine, masculine, non-binary, etc.)?” outline that AI has an “undefined” gender (76/109 respondents, that is 69.72%). We also note that a more masculine, not feminine, representation of AI dominates, in particular among high school students. These results are to be linked with the gender perception (masculine) of the developers of generative AI tools. Indeed, “AI reflects the biases of those who dominate this conception and triggers discriminations that are the reflection of the biases our society holds” (Nurock, 2019, p. 69).
- 11 Only the keywords cited at least twice per respondent have been kept.
18In response to “How would present AI in five keywords?,” the most quoted keywords11 are divided as follows:
-
Technological aspects: algorithms, calculation, ChatGPT, Google, data processing, intelligence, robot, and technology;
-
Aspects related to knowledge, training, and its utility: adaptation, training, autonomy, knowledge, creation, data, convenience, information, tool, research, learning, simple, and useful;
-
Aspects related to perception: future, danger, scary, fictive, pointless, and fast.
19Four similarities in representations are notable in the three cohorts:
-
Intelligence/intelligent are the most quoted keywords by the three cohorts;
-
Knowledge and future (or fictive for middle school students) are keywords equally used by the three cohorts;
-
Technology, useful, help, danger (and scary/fear for university students) and fast/rapidity are equally used by university students and high school students;
-
Robot and creation are equally used by middle and high school students.
20Divergences are also to be underlined:
-
- 12 Baromètre 2025, “Les Français et les IA génératives – Vague 3,” Ifop pour Talan, March 2025. https: (...)
The keyword “information” is used only by high school students. This result may be linked to the result presented in the Ifop report which indicates that the primary usage of a generative AI tool is “to undertake research as in a search engine;”12
-
- 13 Ipsos report, “L’usage de l’intelligence artificielle par les Français,” February 2025. https://www (...)
The names of generative AI tools are only used by middle school students. ChatGPT presents itself as the primary generative AI tool used by professionals;”13
-
- 14 As Fred Cavazza notes: “AI for AI’s sake is useless, what is important is to define instances of us (...)
- 15 Such as the Unesco in “Generation AI: Operators or innovators?” which states that “one can then ima (...)
Middle school students (despite the low level of respondents) quote at 92% the word “data processing” among the keywords. They also underline two contradictory aspects: for some middle school students, generative AI may be pointless (“it’s useless”14) while others say that it is conducive to learning;15
-
Finally, students present generative AI in their keywords as having the quality of being simple and able to adapt, thereby facilitating training and autonomy through the provision of simple and going-forward research tools (algorithm, calculation, data). Concerning the latter point, for Anders and Spelz (2025), the rise in “self-efficacy” is a powerful marker of autonomy. The results of their study show that students feel they can use AI without being completely dependent on constant support. They are therefore encouraged in their own learning process.
21Four questions were asked to respondents in this dimension. Their answers are divided as follows:
221. How do you picture the future impact of generative AI on your personal life on a scale from low (1) to high (10)? The responses reveal that high school students think that generative AI will have a higher impact on their personal life in the future (27.7% between 8 and 10) than the other two cohorts (21.42% on scale 5 for university students). These results are to be linked with the fact that AI should not simply be evaluated according to criteria of technical performance or profitability, but also according to its effects on the well-being of individuals and of its communities (Schiff et al., 2020).
232. How do you evaluate the future impact of AI on your professional life on a scale from low (1) to high (10)? The responses lead us to stress that the impact is similar for the three cohorts (more than 50% selected the higher numbers on the Likert scale), with an even higher importance among university students (82.96% on scales 8, 9 and 10). These responses highlight the need to roll out among university students generative AI competencies for a successful professional integration (Moujtahid, 2025). This need is also to be linked with the feeling of noncompetency underlined by European citizens (Scantamburlo et al., 2023).
243. How do you evaluate the future impact of AI on the environment on a scale from low (1) to high (10)? The feeling among university students is divided between “average impact” (49.11% on scales 4 and 5) and “very high impact” (36.16% on scales 9 and 10), whereas the majority of the other two cohorts (middle and high school students) see the impact on the environment as “average” (scale 5). These numbers are to be linked with the data collected on the notoriety and image of generative AI in France16 (2019) which reveal that 76% of French people think AI will play a major role in the environment and the energy sector in the next few years.
254. How do you evaluate the future impact of AI on society on a scale of low (1) to high (10)? The representation among middle school students appears “very high” (37.71% on scale 10), as well as among university students (87.21% on scales 7 to 10). It is lower among high school students (69.99% on scales 7 to 9). This result correlates with the research published in 2022 by Littman and his colleagues (2022) which highlights the possible consequences of AI, namely social disparities, concentration of power, unfulfilled ethics, and to-be-developed transparency.
26In this last dimension, the responses to the question “Have you used an AI tool? If yes, which one and how often?” are differentiated. From middle school onwards, generative AI tools are known and used by most respondents, reaching 100% for university students. The best known is, without surprise, ChatGPT, with an increasingly important diversity of generative AI from middle school to university. The frequency of use increases with the educational pathway (from several times per year for 78.57% of middle school students, to everyday for 31.91% of university students). These results are similar to those proposed by the Baromètre 2025 report which notably specifies that ChatGPT is currently the most used generative AI tool in professional contexts.
27Finally, the different item responses (free in the questionnaire) to the question “In what context did you use these AI tools?” are divided into five categories (115 items in total):
-
The omnipresent use of AI tools in academic activities from middle school to university, even if the number of items, it must be noted, is smaller among middle school students than the two other cohorts;
-
High school students use more AI in computer research than the two other cohorts;
-
The use of AI for leisure and entertainment is important for middle school students and, to a lesser extent, for university students;
-
Middle school students say they do not use generative AI out of curiosity, contrary to high school and university students;
-
Only high school students formulate their usage of generative AI in terms of orientation, according to two axes: “to find graduate schools to list in [French access platform to higher education] Parcoursup and for research on course contents,” “to choose the subjects specific to my second year in high school”. This result may be linked to a study led in 202417 which stipulates that 99% of high school students use generative AI and 92% use it regularly. There are even 30% of them who use it daily.
28Recent advances, in particular through models of natural language processing, in vision systems and autonomous agents, have made generative AI technologies accessible to a large public. If their potential is considerable, their mass deployment raises concerns with education and training. In this context, we chose to question the representations of middle school, high school and university students, with the view to helping them build up a personal professional pathway.
29In terms of the relationship between generative AI and everyday life, our results reveal a non-perception of generative AI’s biases from the respondents, which is causing the development of already held discriminations in our societies. Examining societal biases (as they present themselves on the web, for example) and their various implementations in generative AI tools could lead learners to question the validity of different sources of information.
30In terms of the definition and perceptions of generative AI in different contexts, we note the need to develop the skills shared by the three levels of the study’s participants. But we also highlight the need to identify the generative AI skills specific to each level (middle school, high school and university) and each training-targeted job, as well as the need to imagine the development of new skills for jobs that are not yet known.
31Finally, as far as the dimension related to the usage and utility of generative AI is concerned, the results reveal that it is used for educational purpose, particularly regarding the search for information (like a search engine) for the majority of the three profiles of respondents. High school students being the only group to use it for their orientation, further consideration should be given to orientation programs and their training methods to make each learner more autonomous in the management of their orientation. It would open up methodological perspectives on the job market, identifying fast-growing job tendencies in certain geographical zones by proposing a series of factors (salaries, conditions of employment, etc.), thereby providing the basis for a new supportive approach to academic and professional success (aim of the programs “Cordée de la réussite”).