Only 17% of French knowledge workers report a good relationship with their work in 2025, down from 21% the previous year according to the 2025 edition of the HP Work Relationship Index (WRI). This worrying decline occurs even as the use of artificial intelligence in companies continues to rise. The paradox? AI significantly boosts job satisfaction (+11 points among respondents who confirm using it), but access and usage vary from one employee to another. This digital divide raises a triple challenge for organizations: equity among colleagues, the collective performance of the organization, and data security. A unique window of opportunity opens to democratize AI access through a renewal of IT estates.
Growth of Inequalities and Security Risks
The figures from WRI 2025 highlight a first finding: in day-to-day practice, 34% of business leaders use AI tools provided by the company, compared with only 13% of knowledge workers. This gap is doubled by a skills gap: 46% of leaders master AI, versus barely 17% of employees. Even more troubling, access to training is retreating for those who need it most: 36% of knowledge workers* received AI training in 2025 (down 5 points in one year), against 60% for leaders.
We therefore observe a two-speed AI: one driven by management, and one available to staff. The consequences are clear: while the arrival of AI is driving a profound transformation of the world of work, 83% of knowledge workers* in France find themselves affected in their relationship to work.
Yet the WRI 2025 barometer shows that when employees have the right technological tools, their chances of having a good relationship with their work are doubled—and even quintupled when they see that their company invests in them.
Beyond employee well-being, access to AI tools is also at the heart of security questions: in the absence of professional tools provided by their company, 85% of French employees report using AI in their tasks via solutions they choose themselves (ChatGPT, Midjourney, …) without approval from their IT department. This phenomenon, known as “Bring Your Own AI” (BYOAI), exposes companies to major risks: data leaks, security breaches, GDPR non-compliance, loss of process coherence, etc. The classic Shadow IT is thus complemented by an uncontrolled Shadow AI. This usage fracture is therefore not just an HR problem: it is a direct drag on competitiveness and a risk to organizational security.
2026: Towards Mass Adoption of AI
Companies face a mandate: continually upgrade their IT estates to meet employee expectations as well as the technical realities and rising cyber threats. By opting for AI-powered PCs, IT decision-makers can address several issues at once: increasing productivity, preserving the integrity of company data, and optimizing employees’ time. According to HP’s internal forecasts, 60% of IT leaders expect to purchase AI PCs within the next 6 to 12 months. IDC’s global study confirms this trend with a market surge: from 50 million units in 2024 to 167 million in 2027, representing 60% of global PC sales. IDC notes that the modernization of operating systems will be the main catalyst, with a commercial renewal peak expected in 2026.
What really changes the game? Local AI processing. The new generation of AI PCs includes a Neural Processing Unit (NPU) dedicated to machine learning, delivering up to 55 TOPS of compute power courtesy of the AMD Ryzen AI PRO processor found in devices like the HP EliteBook Ultra or the OmniBook Ultra. The benefits are immediate: no latency, guaranteed data privacy, and access to AI even offline. A direct countermeasure to BYOAI.
AI PCs provide a solid technical answer (embedded, local, secure, transparent AI), but hardware alone is not enough if the software side is not optimized. At HP, AI is envisioned as a true professional assistant: HP Smart Sense automatically optimizes performance and battery life based on activity. HP AI Companion learns the user’s work habits and automates repetitive tasks, without requiring any special technical expertise.
The Hardware, a Lever for Democratizing AI
Trust in AI remains fragile and complex, as revealed by the global KPMG Trust in AI study conducted with 48,000 people across 47 countries, including France. If 66% of respondents use AI regularly, only 46% trust AI platforms. More worrying: 56% make work errors because of AI, 66% rely on results without verifying accuracy, and 70% think regulation is necessary.
And if IT decision-makers are the key to AI adoption? According to the WRI 2025, 71% of them believe they can have a significant impact on the employee experience (EX) in technology adoption. Yet they are only 37% to be involved by leaders in EX projects—a surprising paradox showing that those who could bridge the digital divide are often excluded from strategic EX decisions.
But equipment is only the first step. The real democratization of AI requires structured support: training tailored to different skill levels, awareness of best practices, and progressive integration into work routines. The more employees are supported, the better they will understand AI—and the more effectively and confidently they will use it. Inclusive AI does not start solely with IT equipment; it requires a combined approach of equipment x training x support. IT decision-makers, the architects of this transformation, play a decisive role in narrowing the gap and building AI that is truly accessible to everyone. On the condition that their role is fully recognized in this AI odyssey.
Want to delve deeper? See the full HP Work Relationship Index 2025
*Knowledge workers, or knowledge professionals, are workers who primarily mobilize cognitive, relational, and communicative abilities, in collaboration with other workers and/or with machines in the context of their professional activity.
