Artificial Intelligence Usage Guide

The availability of these tools and their ease of use undeniably has a considerable impact on all higher education and research institutions. Inalco is particularly concerned, as these tools offer language generation ("genAI") features, whose capabilities however vary enormously depending on the language.

Over the past few years, AI tools have been developed and gradually made available to the general public. The release of ChatGPT by OpenAI at the end of 2022 was enormously successful, well beyond the tech and developer communities, particularly among younger generations.

Definitions

AI

Artificial Intelligence (AI) algorithms aim to reproduce human cognitive abilities through calculations performed on an electronic device (computers).

Generative AI

Generative AI models use algorithms to determine which words are most probable in a process of artificially generating data (texts, images, videos, etc.). Generation is initiated by a "prompt" (a command written in natural language). The range of use cases is broad.

Large Language Models

Large Language Models (LLMs) are computational models built by adjusting a very large number of parameters calculated through artificial "learning", based on very large volumes of text. They model language by identifying regularities in textual data. Their parameters and statistical calculations allow them, among other things, to "generate" seemingly coherent texts, without it being possible to claim that they "understand" texts the way a human does. These models can be developed for a single language or for multiple languages simultaneously, depending on the textual data on which they are trained. The performance of these models varies greatly depending on the language: languages with a strong online presence (English, French, Spanish) yield better results than digitally under-resourced languages.

Uses of AI

As these tools can be used for many purposes, it is useful to distinguish between them in order to better identify the methodologies to adopt for each use case, and to determine the risks and frame appropriate practices.

In the context of work carried out at Inalco, the following uses can be listed:

• information retrieval,
• identification of research questions,
• structuring of ideas (e.g. in the form of an outline),
• proofreading and suggestions for improving a text,
• automatic translation of texts,
• digitisation or transcription (images for writing or audio files for speech).

General-purpose AI tools offer to perform these tasks, though the quality of their output remains highly variable depending on several parameters (task, language, input data, etc.). There may also be specialised AI-based tools for each use case; we do not provide a list here, as the available tools are numerous and evolving very rapidly.

It is important to note that these AI tools are progressively being integrated into existing applications where the use of AI is not always explicitly mentioned (search engines, text editor spell-checkers, document summarisation tools, videoconference transcription, etc.).

Recommendations for Good Practice

When using AI tools, it is essential to review what the tool has produced with a critical eye and a great deal of vigilance. Numerous cases of "hallucinations" have been observed (when AI makes claims based on no verifiable source), particularly in specialist fields and when generating bibliographic references. When AI is used in a piece of work, it is recommended to declare this explicitly and to indicate which parts of the work involved AI tools (data analysis, information retrieval, translation, writing improvement, etc.).

If the use of AI is being considered, it is strongly recommended to do so in an assistive and suggestive capacity, and not to rely on it to generate the work itself. Indeed, an AI can offer a perspective on a piece of academic work. The author may take note of the suggestion and act on it if they find it relevant.

We invite those who use AI to be particularly vigilant about the potentially biased or incomplete nature of generated content. Such content is indeed designed to provide answers based on sources selected by AI manufacturers, which are often partial and incomplete, and these tools very rarely acknowledge their limitations or lack of knowledge, particularly in fields requiring deep expertise. Furthermore, their use in the context of research carries the risk of confining that research to the directions suggested by the AI; it is therefore strongly recommended to use them sparingly when starting a research project.

Several studies have shown that certain uses of AI can be detrimental to the acquisition of knowledge and skills. We recommend that everyone use this technology in a considered manner, so as to make AI a useful learning aid that complements rather than replaces traditional learning methods. Furthermore, the use of AI in the context of Inalco activities should, as far as possible, be consensual — you may contact the AI Mission (link below) if you observe inappropriate pressure to use AI. Finally, it is widely known that AI tools are extremely energy-intensive; their use and the calling of their features should therefore be considered carefully and, where possible, weighed against the energy consumption they entail.

Prohibitions

For Inalco and in a pedagogical context (coursework and examinations), students are asked to respect the following three prohibitions:

• do not submit personal or confidential data to an online AI tool,
• do not present AI-generated text as one's own personal work,
• do not use automatic translation in language coursework, unless such use is explicitly authorised by the instructor.

These uses of AI are currently explicitly prohibited under the internal regulations (Article 71) and the assessment regulations (Article 2.2), and are subject to disciplinary sanctions.

Page last updated on 27 March 2026.