The season of wishes is in full swing and, as usual, everyone benefits in these first four weeks of the year from an almost unlimited supply of wishes and all kinds that allow us to hope for the best, sometimes even if he means fall. in utopia. A month before the future summit devoted to artificial intelligence (AI) which will bring together States and NGOs in Paris on February 10 and 11, let’s take advantage of this magical parenthesis where all hopes are allowed to express some hopes for the future . of AI.
AI superstar
This challenge is even more necessary, since at the rate at which AI advances, it is a safe bet that this year 2025 will bring a number of significant advances that will make this technology even more significant. Whether it is the “agentic” revolution through the implementation of generative AI in everyday activities through automatic programs (“agents”) that will become true co-pilots of life, or the advent (“agents”) very soon “according to Sam Altaman, head of Open AI) of AGI (Artificial General Intelligence) able, we are promised, to perform any cognitive task that a human can perform and therefore competition with. human intelligence, it should not be wrong to predict that the coming months should see the emergence of a myriad of debates and questions centered on the place of AI in our daily lives, and especially those on to the “Men-Machines” cohabitation.
Between hopes and fears
At the bottom of these hopes and fears, the fact that these technologies can help solve many problems facing humanity (global warming, scientific research, etc.) while highlighting the risks that they gain the ascendancy until to the point of gradually losing ours. status as “masters” and become only “parameters”. As is often the case, the truth lies somewhere between these two extremes, bearing in mind that in the current AI “boom” phase, the prophets of these technologies also have the interest to promise progress necessarily revolutionary, although – this is to create demand while focusing on the continued appeal of this sector to investors.
A wishful thinking?
However, the period in which we find ourselves, where AI is starting to enter entire sectors of our economies and our lives, is undoubtedly an appropriate moment to think about the AI we would like for the coming years . Far from a simple techno-critical approach or alarmist discourses of the science fiction type and without, and vice versa, falling into ecstatic remarks, the wish that should be formulated would be that this future summit of AI manages to define the characteristics of an AI that could be qualified as “of general interest”. Many thinkers have already made a sketch of this technology that should be at the service of everyone.
For the science fiction author Alain Damasio, in his latest work “Vallée du Silicium”, he must ” ensure that our technologies are or become user-friendly, thus rebranding AI as “Friendly Intelligence”. This would take the form of a technology that emancipates those who use it, that frees cognitive abilities… in short that expands the personal action of everyone.. » Daron Acemoglu, Nobel Prize for Economics 2024, extends this idea by expressing hope:
” that we need to work towards AI that serves humans… To achieve this, we need a new narrative in the media, in political circles and in civil society, as well as better regulations and policy measures.. »
Digital Commons
One of the possible variants of this AI at the service of humans, or the term “AI of general interest” could be based on the creation of “AI-specific digital commons, i.e. shared corpora and /or open-source”. so that the datasets essential for the creation of AI for scientific, research or educational purposes can see the light of day and be shared in complete transparency.
This project to create digital commons specific to AI would be almost the sine qua non condition to guarantee equitable access to the resources necessary for the development of AI of general interest. Whether it is open data sets, shared models or algorithm libraries, such “commons” would be the guarantee of a collaborative basis that allows us to share knowledge useful to all and to democratize innovation without forget that they also help to reduce the concentration of technologies. power in the hands of a few large technology companies, ensuring transparency and trust of users and designers. Wanting these “AI-specific digital commons” would be laying the foundations for a truly inclusive and ethical AI.
Global AI governance to be invented
Alongside this idea of a “shared digital commons” that will significantly enable the development of new open models of AI, it could also be relevant to consider the global governance of AI. It is not until a cataclysm, the Second World War, that international institutions (UN, UNESCO, etc.) emerge as places of debate and pacification of conflicts. On the subject of AI, especially when the hypothesis of super-powerful machines are regularly mentioned, even sometimes by those who develop them, it would be time to imagine this future global governance of AI.
Such a project provides the means to prevent abuse while trying to get the most out of this technology. This international framework to be invented could, for example, be envisaged via a dedicated agency, based on the model of the International Atomic Energy Agency (IAEA), to monitor and regulate innovations in AI while establishing standards ethical, technical and security, and this to participate in States, companies, researchers and NGOs.
We will recognize that the challenge of these new rules of the desired game is immense, even utopian, but at the beginning of the year when all wishes are possible, do not formulate the wish for this AI that we would like to be a failure, even a mistake
