Google/Ipsos bva study on AI: adoption still cautious in business

One in two French people used artificial intelligence in 2025. Yet in the workplace, adoption remains hesitant, uses are often basic, and training is almost non-existent. A large Ipsos bva survey for Google provides a comprehensive overview and highlights levers to accelerate adoption.

Adoption is accelerating, but understanding remains limited

France has caught up somewhat on consumer AI. In two years, the number of users has almost doubled.

France ranks ahead of the United States (40%), Japan (42%) and Germany (47%), but lags behind the United Kingdom (56%), Italy (59%) and Spain (64%).

Using AI does not mean understanding it. On this point, the French struggle more: 59% say they have a good understanding of what AI is, versus 66% in the United States, 64% in the United Kingdom and 75% in the Netherlands. The survey points to a cultural explanation: in France, AI is discussed more from the perspective of regulation than concrete uses.

Understanding gaps also vary greatly by profile:

These disparities are not trivial: they are reflected, and amplified, in professional uses.

An ambivalent perception: beneficial for the economy, worrying for society

The image of AI has improved. Nearly one in two employees (46%) now has a positive perception, while only 22% remain negative.

But the perception remains fundamentally dual. Employees see AI as an opportunity for companies (41%) and for themselves personally (37%). By contrast, they are much more skeptical about the collective impact:

When asked about the main potential drawbacks, employees cite first excessive dependence on the tool (25%), then loss of human interactions (24%), ahead of risks to data privacy (23%) or the fear of being replaced by AI (21%). Fear of technological unemployment is therefore far from the primary driver of resistance.

Executives, however, are convinced

Senior executives have a radically different view. Asked what will have the most positive impact on their organization in the next five years, 53% cite advances in AI as their top priority, far ahead of the ecological transition (28%) or improving employee skills (23%).

Their enthusiasm is based on concrete results:

• 70% estimate that AI has already improved their company's productivity
• 52% say it has saved employees time — more than 3 hours per week in two-thirds of cases
• 76% have reallocated that freed-up time to higher-value tasks
• 64% have trained their employees in new skills
• 57% have increased employees' salaries following the AI deployment
• 75% have even hired new staff to support this deployment

Far from the image of a technology that would destroy jobs, leaders are not very worried: only 5% would consider cutting staff if jobs could be assigned to AI. And 80% consider it likely that their company will use these tools more in the coming year.

Within teams, usage remains too timid

Despite this optimism at the top, actual use within teams remains limited.

Among those who use AI at work, tasks remain mostly basic:

More advanced uses are mainly seen among younger people. Among those under 25, 30% use AI for creativity, 26% for training and 23% for technical tasks, compared with 24%, 13% and 13% respectively among 55-64 year-olds.

Why do these uses remain so limited? The main barrier is simple: 27% of non-users believe AI is not relevant to their tasks. Lack of interest comes next (19%), followed by doubts about the quality of results (15%) and difficulty integrating it into work processes (13%). Lack of awareness about the tool's potential is therefore the real bottleneck, far more than technical or ethical reservations.

The "Shadow AI": a sign of genuine appetite

A revealing phenomenon emerges from the survey: the "Shadow AI".

42% of employees who use AI at work access it mainly via their personal account, compared with only 29% who use a professional account provided by their employer. And only 14% of workers say their company has implemented an internal policy governing this use.

Senior executives are aware of this: 35% observed that employees used their personal accounts for professional purposes, and 37% noted that employees had started using the tool even before any official executive was designated.

This phenomenon is certainly a data privacy risk. But above all it reflects a real desire of employees to explore the potential of these tools, even without being encouraged or trained. A signal employers would do well to heed.

Small businesses and sole proprietors: structurally behind, but pioneers in usage

Small businesses account for more than 99% of French companies and more than a third of private-sector jobs. Their lag is therefore a key issue for the country's AI transition.

Yet, individually, leaders of small firms fare better than the average employee, both in frequency of use and in quality of use.

They are also more convinced of the benefits: 44% of micro-business leaders and 41% of self-employed entrepreneurs believe AI can improve their efficiency, compared with 33% of employees. The gap is just as marked for creativity (33% and 32% versus 22%).

These profiles often discovered AI on their own, self-taught, and have turned it into a concrete business development tool: brand identity creation, cost-price calculation, strategic planning. Reservations nevertheless remain in some sectors — crafts, catering, manual trades — where the link between AI and daily activity is still hard to perceive.

Training: the most powerful lever, and the most underutilized

The figures on training are among the most striking in the survey.

The impact of training is, however, massive. Trained employees use AI at least once a week at a rate of 68%, compared with only 26% for those who have never been trained. That's a ratio of 1 to 2.6.

Trainings also have a direct effect on confidence and practices:

• 76% of those trained identified new use cases thanks to the training
• 74% have gained confidence in using AI
• 74% have started using AI for new tasks
• 67% have seen their AI-related fears decrease

Paradoxically, 58% of executive managers say they have implemented AI training in their company, a figure far from the 21% of employees trained. These trainings, when they exist, likely do not reach all employees or are not perceived as sufficiently relevant.

Employees' expectations are clear: 62% want training supported by concrete case studies, 61% want practical training on how to use the tools. Demand for theoretical and abstract content is much lower (54%). The issue is not to explain "what AI is," but to show "how to use it in my job, day to day."

To deliver these trainings, employees trust technology companies (27%) more than employers (24%), training organizations (22%) or educational institutions (18%). Small business leaders and self-employed entrepreneurs turn even more to tech players (34% and 32%).

Finally, AI skills are becoming an increasing hiring criterion: 58% of French executive managers say they have already hired a candidate partly for their AI skills, above the European average (54%). For employees, mastering these tools is no longer just an asset: it is becoming a condition of employability.

The article "Google/Ipsos bva study on AI: adoption still cautious in business" has been published on the site Abondance.