Invited on the occasion of the next AI France Summit 2025, Jean-Gabriel Ganascia, President of the CNRS Ethics Committee, addresses the concrete uses of artificial intelligence. Companies are currently making positive use of it, provided they respect the right dosages.
The uses of artificial intelligence, particularly by businesses, are now anchored in reality. Do you share this evolution and analysis?
Jean-Gabriel Ganascia : It seems to me that we still wrongly consider that artificial intelligence was not mature. And this has been the case for several years. Many people forget that many technologies that have been used for a long time come from AI. I am thinking, for example, of pattern and voice recognition for dictation services. These tools have already been present in companies for many years. I am also thinking of object-oriented language, part of which finds its technological source in AI recognition tools, or even expert systems.
For its part, generative artificial intelligence is proving very impressive. This leads to projections about its development in the future. I would even qualify some speculations as very ambitious. Some of them will come true but others have little chance of succeeding. but I discover many interesting and relevant new applications every day. So the potential is there.
So you don’t believe in an exponential development of AI in the coming years?
Jean-Gabriel Ganascia : Experts, just like companies, realize that the results produced by artificial intelligence, including generative, can create many hallucinations. The latter will then engage the responsibility of the institution or company that uses it.
The results produced may be highly speculative. And some artificial intelligences behave like parrots by stringing together words without confronting them with the reality of the facts. It is then possible to use RAG (Retrieval-Augmented Generation) to specialize the responses produced and rely on a reference corpus. But then the machine does not tell the whole truth.
Many investments made so far risk disappointing in the long term. I therefore recommend caution to everyone because everything we foresee is far from certain to come true.
However, artificial intelligence, particularly generative, is expected to become widespread in certain areas and offer major uses, particularly in the field of translation. If we only trust technology, then dangers can quickly arise with the risk of misunderstandings.
Translators will still be able to proofread the work provided by the AI, but their task will be tedious. Technology could therefore create deleterious effects. If the idea is to overcome language barriers, it is also likely that artificial intelligence will not fully achieve this goal.
-Numeum highlights the notion of pharmacological use of technology. Do you join this position?
Jean-Gabriel Ganascia : I completely agree with this position. Technology constitutes both a drug and a medicine. This is where the difficulty of using technology lies. In the educational world, for example, one of the major topics is teaching students to write correctly. But this can’t just be learned in class. Also, teachers can use artificial intelligence and lead to formidable simplifications. Whereas we expect the opposite of commonplaces from school. We have to bring subtlety to it.
In short, the risk is not that students cheat but that they no longer know how to do without it. For its part, the University is made to acquire knowledge and develop. It is therefore not impossible to see cases emerge again in the coming years in which generative AI will be prohibited.
A part of the population has developed a certain fear of AI in the past. Why did this feeling disappear with its generative aspect?
Jean-Gabriel Ganascia : The particularity of generative artificial intelligence is that the population has the feeling of touching it with their finger. It interacts with it in the form of a dialogue, whereas previously, the AI could appear obscure, capable of taking the place of an individual.
But this fear has not disappeared. It continues to exist but in another form.
Is it still reasonable to believe in progress, particularly technological progress?
Jean-Gabriel Ganascia : The danger with the field of artificial intelligence is that actors dominate. They even continue to expand their power. Their predominance risks expanding, like what we are seeing with Elon Musk in the United States. My fear is therefore not technological. The latest Nobel Prize in Chemistry for the particles of living things is a clear sign that technology can better care for the population. By creating new drugs using AI. This opens up extraordinary perspectives but it depends above all on the men behind this technology.
My concerns are therefore more political than metaphysical. For example, the problem is not to create false information or to profile it to disseminate it to certain people in particular to add fuel to the fire and encourage certain categories of the population. The risk is then to drown Internet users with information which creates an extremely strong and even suspicious climate.
The AI France Summit 2025 exhibition once again brings together all artificial intelligence experts and professionals this year. Organized by Numeum, the benchmark cross-sector event highlights technology as a concrete reality for all companies in the ecosystem, including digital. This year’s theme is ” Electrify the Now of AI: Together We Create“. The idea is to transform the vision of artificial intelligence into a reality that propels everyone’s ambitions to new heights.
Olivier Robillard
