In June 2025, according to Crédoc’s Digital Barometer, 48% of French people aged 12 and over had used generative AI during the year, up from 20% in 2023.
This rapid growth doesn’t tell the whole story: in the same survey, 18% of users said that a platform they already used had imposed a generative AI feature on them, without them choosing to use it. Technology is no longer simply responding to expressed needs. It is reshaping digital habits before users have even made a choice.
Generative AI adoption in France: less spontaneous than it seems
The rise from 20% to 48% of users in two years suggests widespread adoption. Yet significant differences remain: younger people use generative AI far more often than older groups, while education level and occupation also have a strong influence on how frequently it is used.
What stands out is the share of adoption driven by default settings. When a search engine, messaging service or social network adds a layer of generative AI to its interface, users aren’t always aware they’re using it. 18% of users didn’t choose the AI feature they use: it was simply there, enabled by default.
This raises a question rarely asked in discussions about digital transformation. Does usage reflect a genuine need, or the ability of platforms to make a feature seem necessary simply by placing it in the interface? Coverage from technovox.fr offers a clearer picture of how these technological trends are spreading into everyday life.

Who decides how we use digital technology tomorrow: users or platforms?
Adding AI features by default to existing services blurs the line between voluntary adoption and passive exposure. A messaging app user who sees automatically generated reply suggestions hasn’t asked for them. They may end up using them simply because the button is there, in the right place at the right time.
This pattern isn’t new. Social networks did the same with algorithmic feeds, replacing chronological displays without consulting users first. The difference with generative AI lies in the scale of the change: it’s no longer just the order in which content appears that is changing, but the very content presented to users.
The role of interface design in adoption
Design choices shape behavior. A feature enabled by default will be used by most people, while one that must be activated manually will remain marginal. This bias, documented in cognitive ergonomics, takes on new significance when applied to tools that can write, summarize or rephrase content on a user’s behalf.
Default settings shape usage more than expressed needs do. Platforms that roll out generative AI on an opt-out basis (requiring users to turn it off) rather than an opt-in basis (requiring users to enable it) mechanically generate high adoption figures, without those figures reflecting a deliberate choice.
European AI Act: transparency requirements from August 2026
From 2 August 2026, the European AI Act will introduce specific requirements that directly affect everyday use:
- Users will have to be notified when they are interacting with certain AI systems, so they know whether they are communicating with an automated system or a human.
- Certain content generated or manipulated by AI will have to be clearly labelled, particularly images, videos and texts that could mislead the public.
- Providers of general-purpose AI systems will have to document their models’ capabilities and limitations, making external oversight possible.
These requirements will take effect on 2 August 2026. Their implementation will change how digital content is presented in news feeds, search engines and communication tools. For users, this is the first time a European regulatory framework has required platforms to make visible what has, until now, been invisible by design.
The real impact of these measures will depend on how they are enforced. Global platforms will have to adapt their interfaces for the European market, requiring technical and economic trade-offs whose outcomes are not yet known.

The energy footprint of data: the material cost of digital use
Everyday digital transformation depends on physical infrastructure that remains largely invisible to end users. In France, data centers consume around 10 TWh of electricity per year, or 2.2% of national consumption, according to ADEME figures. Some services used in France run on data centers located abroad, so the true footprint of digital use extends beyond the country’s infrastructure.
Local tensions over data centers
Data center projects in France have faced opposition over their location and resource consumption. These protests show that the digitization of everyday life isn’t just about apps and interfaces. It is also a matter of land use, access to water and electricity, and coexistence with existing activities.
Digital transformation has a material cost that its interfaces make invisible. A voice assistant that responds in a second relies on servers, cooling and energy. The gap between the perceived lightness of using the technology and the heavy infrastructure that supports it is a blind spot in our collective understanding of technology in everyday life.
Technology and everyday life: what adoption figures don’t tell us
Usage statistics measure what people use, not why or how. A 48% adoption rate for generative AI doesn’t distinguish between someone actively writing prompts and someone clicking an automatic suggestion without giving it much thought.
This distinction matters when anticipating how people will use technology in the future. If adoption is mostly passive, the features that take hold will be the ones platforms choose to enable by default, not the ones users ask for. The future of digital use depends as much on default settings as on technical innovation.
The European AI Act, changes in behavior measured by Crédoc, and tensions over infrastructure paint a picture of technology reshaping everyday life according to forces that are partly beyond the control of those who benefit from it. The question is no longer whether technology changes our lives, but who sets the terms of that change.



