Advocating for a No-IAg label in education, culture, journalism, research and businesses

Ten years ago, faced with the growing importance of pre-cooked meals in catering, a small, unpretentious logo emerged, without a certifying body, and without a labeling process: the “homemade” label. A sign of trust and recognition between a restaurant and its customers, it indicated that – as should go without saying – the food that one was about to eat had been cooked on site, and not simply reheated in the microwave after having been produced in a factory.

Ten years later, the surge of generative artificial intelligence (AIg)[1] in all aspects of daily life disorients the many people who are attached to human productions and suffer this wave of dehumanization with fear, sadness or anger: no longer possible to know if a newspaper article, a piece of music or educational content was produced by an AIg. On the creator side, the situation is not much rosier: faced with competition from ChatGPT and automation, how can you assert and recognize your choice to continue to offer human production?

Beyond the extreme case of productions entirely generated by IAg, the latter can be used to varying degrees during different phases of the creative process: illustrations, funding requests, presentation texts, research bibliographic, or of course assistance with final production. The reasons for refusing all of this outright are multiple and are probably as diverse as the profile of the people who want it. The most common is that AIg is not a neutral tool with its good and bad uses but a technical system inseparable from massive looting of works, exploitation of click workers and gigantic, energy-intensive infrastructures. We can also refuse to leave entire sections of our creative humanity to companies in the sector or refuse to lose resources. skills and autonomy to which we are attached But no matter: whether on the side of the creators or on the side of the people who access content, it is time that we can recognize each other, rally together, trust each other, and create a network of creations. entirely human.

One option would be to ask all people using AIg at any stage of their creative processes to declare it explicitly and in a transparent manner. But we doubt that such a practice will ever become necessary, for many reasons. We can simply illustrate it with an emblematic example: in the field of research, where most journals ask scientists to explicitly declare the use of AIg in their articles, a study estimates that only 1 in 40 scientists using AIg declare it. We are therefore witnessing major and widespread breaches of scientific ethics in this area. It is a safe bet that, in many professions, even if we asked to explicitly display all use of AIg, the latter, which is hardly open and difficult to prove, would remain practiced on the sly.

We therefore propose that a Without-IAg label emerges quickly wherever people, collectives, newspapers, laboratories, training courses or companies make a commitment not to use AIg in their professional practice. This label would be based on trust between humans. It could emerge on very diverse scales. For example, in education, this label could appear at the level of an establishment, a department, a training sector or a teacher. In the media, it could be displayed on the scale of an entire newspaper, a section or a journalist. In culture, it could be highlighted at the level of a work (film, novel, photography, etc.), or to characterize the practice of artists (screenwriters, writers, etc.). It could appear in the form of an explicit sentence, a personal logo or a short text. The important thing is not the form it takes; it Perhaps it is not even necessary that an “official” label and its set of precise – and probably unverifiable – standards exist; the important thing is that all people who recognize themselves in this desire and this commitment display it explicitly, even in the absence of a certifying body or official label, an affirmation. erroneous could be legally considered as a deceptive commercial practice.

Companies are a slightly different case but nevertheless very interesting. We see the emergence of a dystopian world where thousands of CVs generated by IAg are sorted by IAgs to recruit people who will work all day long. All of this makes no sense and requires urgent clarification, so that students can quickly find out which companies or services in which they can exercise their skills with dignity. Conversely, the same companies should be able to quickly recognize students who have real personal skills, because they would explicitly display their attachment to human production in their CV or by the fact of having chosen to follow a course in human training, that is to say without IAg.

It is likely that AIg deployment will not immediately deviate from its current dystopian trajectory. However, we can act now so that the pursuit of a life without AIg does not feel like an obstacle course, and even remains quite possible. We therefore call for collective reflection on this subject in all the institutions concerned. It’s time to choose sides – AIg or human – and clarify your position.

 

This text comes from the inter-disciplinary reflections of Atécopol, a collective of Toulouse scientists committed to the ecological emergency. It was written by Marianne Blanchard, Guillaume Carbou, Julian Carrey, Adeline Grand-Clément, Jean-Michel Hupé, Odin Marc, Sydney Thomas and Laure Vieu.
The editors also took into account comments and suggestions from artists, writers, and journalists who had access to a preliminary version of this text, as well as members of the Scientifiques en rébellion collective.

 

Notes

[1] We focus here on a certain category of artificial intelligence, generative AI, based on large language models (LLM, Large Language Model), which are trained on large corpora of data to generate text, images, sounds, videos.

The original article can be accessed here