Digital technologies are transforming everyday life at an unprecedented pace, but their adoption is far from uniform. The 2025 Digital Barometer, conducted by Crédoc among 4,145 people aged 12 and over, shows that the adoption of generative AI in France varies by age, education level and occupational category. Measuring these gaps helps us understand who truly benefits from these innovations and who bears their costs.
AI energy costs and infrastructure: what the data reveals
The rise of artificial intelligence is not limited to software applications. Every request sent to a generative model uses data centers whose electricity consumption puts pressure on national grids. This strain makes questions about grid connections, planning and cost-sharing between data center operators and other grid users more urgent.
The costs are not shared equally. Areas hosting this infrastructure bear the burdens (noise, land use and pressure on the grid), while the benefits of AI-powered services accrue mainly to connected urban areas. This imbalance deserves to be documented sector by sector.
To follow developments in these technological issues and their practical implications, the official Séries Conseil website offers regular analysis of digital industries and their dynamics.

Data governance and the AI Act: timeline for obligations in France
Europe has chosen to regulate artificial intelligence in stages. The European AI Act has already had its first tangible effects: since February 2, 2025, certain practices deemed unacceptable have been banned. This first wave includes subliminal manipulation and social scoring.
The next deadlines are staggered according to the risk level of the systems being deployed.
| Effective date | Requirement | Parties concerned |
|---|---|---|
| February 2, 2025 | Ban on unacceptable practices; AI literacy requirements | Developers and organizations deploying AI systems |
| August 2, 2026 | Transparency rules for certain AI-generated content | Providers of generative systems |
| At a later date | Requirements for high-risk systems (employment, education) | Companies using AI in these fields |
This timeline shows that digital transformation does not concern technology developers alone. Organizations deploying AI systems in a professional setting also have governance responsibilities. An SME using an AI-powered applicant screening tool will be subject to the same requirements as a large corporation.
Transparency for AI-generated content
From August 2026, content produced by generative systems will have to be identifiable. This requirement aims to preserve trust in information, a matter of direct concern to media outlets, platforms and public agencies that publish content at scale.
AI adoption in France: persistent gaps between groups
Crédoc’s 2025 Digital Barometer does more than measure overall growth. It details disparities that shape access to technological innovations in everyday life.
- Age remains the main dividing factor: young adults report using generative AI tools far more often than older people, who sometimes encounter AI through services they were already using without realizing it.
- Education level plays a decisive role in the ability to identify, understand and use these technologies in everyday life.
- Occupational category influences not only access to tools, but also how they are used: writing assistance for managers, and automation of repetitive tasks in certain technical jobs.
These gaps call into question the promise of technology that is accessible to everyone. Without support policies, digital innovations risk widening existing inequalities rather than reducing them.

Digital health: data quality comes before tools
The digitization of healthcare is often discussed in terms of connected devices or diagnostic algorithms. The most decisive factor comes earlier in the process: the quality of data systems determines how effective all digital health tools can be.
An early-detection algorithm is only as good as the data it uses. If patient records are fragmented across several non-interoperable systems, or if the data is incomplete, the results will be biased. This issue affects hospitals and community-based medical practices alike.
Interoperability of healthcare systems
France has several health databases, but they are only partially interconnected. Healthcare professionals have to juggle software systems that do not communicate well with one another. Until this data infrastructure is consolidated, the promises of connected healthcare will remain partly theoretical.
The real driver of transformation in this field is therefore not the addition of new sensors or applications. It is the organization and governance of existing data that will determine whether health technologies deliver on their promises for the population as a whole.
Innovative technologies: real benefits and access conditions
The technologies shaping tomorrow’s world share one common trait: their value depends less on their technical sophistication than on the conditions in which they are deployed. The AI Act establishes a regulatory framework, but questions around infrastructure funding, user training and data quality remain largely unresolved.
The benefits of digital innovation will not be shared automatically. Every technological advance creates a twofold effect: it makes some tasks easier while requiring new skills. The adoption gaps documented by Crédoc are proof of this. The next step is not to invent new tools, but to ensure that existing ones actually reach the people who need them.



