Having a favourable opinion of artificial intelligence does not mean using it every day or knowing how to assess its answers. This distinction helps to interpret STATEC’s Regards, published in July 2026: 63% of internet users surveyed in Luxembourg see generative AI as an opportunity. The finding describes an attitude towards the technology, not a test of competence.
The study is based on a survey conducted between June and August 2025. Its scope is specific: residents aged 16 to 74 who had used the internet in the previous three months. It therefore does not allow this opinion to be attributed to the entire population, nor does it establish exactly how perceptions stood in October 2026.
The national figure masks a range of experiences
Generative AI refers to tools capable of producing text, images or code, among other things, in response to a prompt. But this shared definition does not correspond to a shared experience. In the study, 72% of 16–24-year-olds see it as an opportunity, while 52% of 65–74-year-olds consider it a threat.
Occupation also matters. Among workers in information and communication technologies, favourable perceptions stand at 84%. Among manual workers, the figure is 47%. These differences describe groups; in themselves, they say nothing about an individual’s ability to learn to use a new tool.
| Population studied | Measured perception |
|---|---|
| All internet users aged 16 to 74 | 63% see an opportunity |
| Internet users aged 16 to 24 | 72% see an opportunity |
| Internet users aged 65 to 74 | 52% see a threat |
Training: a gap worth attention
The most useful contrast when considering support concerns learning activities. Among those who see AI as an opportunity, 68% say they are undertaking training or learning on their own. The figure is 26% among those who perceive it as a threat.
A note of caution about the word ‘training’: the indicator combines structured courses with exploring on one’s own. It therefore does not measure participation in a course alone, much less the attainment of a qualification. Opening a tool to explore it and following a supervised programme are both included here as ways of learning.
Offering support without treating concern as a failing
Our view is that messaging aimed at people who are already enthusiastic risks overlooking other needs. When planning a workshop, asking what concerns people, starting with a familiar task and allowing time to check the answers seems more useful than demanding prior buy-in.
This editorial proposal is not an effect demonstrated by the survey. The results presented do not allow causality to be established: optimism may go hand in hand with learning, but these figures alone cannot tell us which comes first. STATEC itself points out that perceptions can change rapidly.
The right way to use this study, then, is to open up concrete questions about access to learning. The wrong way would be to turn it into a ranking of people who are ‘ready’ or ‘behind’ when it comes to AI.
Sources
- STATEC, Regards 03/26, published in July 2026.
- STATEC, full study: survey scope and results.
- Photograph: Pixabay.
- Photograph licence.
Sources consulted on 1 October 2026.
