How do the French view scientific and technological advances? Universcience researchers have once again looked at the question through the barometer of critical thinking.
For this 2024 edition, the focus is on their perception of artificial intelligence, and necessarily of its generative AI models, omnipresent “in the news since the emergence of ChatGPT.”
The under 34s are more revolutionary
A majority of French people, 61%, consider generative AI as “a major technological revolution on the level of that of printing”. The differences in perception between generations are perceptible.
Among those under 34, more than 70% attribute the status of revolution to GenAI. However, according to those surveyed, this feeling is accompanied by a need for regulation due to the “new ethical and legal questions” posed.
85% of French people are in favor (77% among those aged under 34). Trust has its limits. In addition, they are in the minority in favor of completely autonomous AI solutions. For 68%, AI applications “cannot make autonomous decisions.
The French also perceive limits in the field of creativity and innovation. 59% consider these solutions “neither creative nor innovative”. Reason: they assemble already existing content. Reservations also relate to their reliability.
Reliability widely discussed
Only 19% believe them to be perfectly reliable. The majority, 69%, say the opposite. These questions about the reliability of IAGen undoubtedly naturally influence the nature of the tasks entrusted to it: detecting errors (75%), analyzing large quantities of data (73%), finding information (70%) or write a text (64%).
On the other hand, “distrust dominates when the question of responsibility comes into play, whether it involves rendering legal decisions (25% confidence), piloting a plane (28%) or a car (34%) , or to prescribe medication (33%) or to carry out a medical diagnosis (41% confidence).”
On the neutrality of these artificial intelligences, and therefore ultimately on the problem of bias, the French responses are more nuanced. Are AIs “more neutral than humans”? “42% answer in the affirmative, 38% the opposite and a significant proportion (20%) do not comment.”
In truth, the responses of these technologies are the result of the data provided during training and the moderation parameters set by the designers.
More risks than benefits: to be demonstrated
Regarding the effects of the development of AI on different sectors, nuance is also required among those surveyed. Overall, they believe that AI presents “as many risks as benefits”.
“On the other hand, it is the perception of risk that dominates for the field of human relations (37%), compared to 31% “as many risks as advantages” and 10% “mainly advantages”),” note the authors.
The results are similar with regard to information (33%) and employment (31%). On health (29%), research and science (28%) and transport (26%), the benefits prevail. However, it is still the nuanced vision that predominates (“as many risks as benefits” at 35, 36 and 33% respectively).
Digital companies do not have the trust
On this question of risks and benefits, the barometer highlights differences between CSP and generations. The CSP+ are more likely to see benefits that outweigh the risks.
Among 18-24 year olds, “the perception of the benefits of AI frequently prevails more clearly”, particularly in the area of transport and mobility. Paradox? Generative AI solutions are today mainly developed by companies, particularly American ones.
However, when it comes to assessing risks and benefits, the French express a strong distrust of the digital giants. Only 29% trust them. Only elected officials and political representatives do worse at 22%, far behind scientists (68%).