Key takeaways:
AI usage is advancing faster than skills development, creating a gap between what the tools make possible and what organisations can genuinely do with them.
Unstructured AI adoption generates organisational debt: tool proliferation, fragmented practices.
Managing AI skills requires clear governance: who uses AI, for what purpose, and with what level of autonomy?
The expected competency goes beyond producing an AI output: interpreting results, identifying limitations, and taking responsibility for them.
Lasting AI integration rests on alignment with existing skills frameworks, decision-making processes, and established ways of working.
Artificial intelligence is gradually taking hold in French organisations, but its adoption is revealing a growing imbalance between usage and genuinely mastered AI skills. Faced with this reality, the question is no longer whether to integrate AI, but how to do so without creating organisational debt.
Multiplying tools, experimenting without governance, or decoupling AI from existing practices exposes projects to stagnation. By contrast, an approach grounded in critical competencies and the progressive development of know-how makes it possible to embed AI on a lasting basis.
AI Is Advancing Faster Than Organisational Skills in France
What the Figures Say About AI Adoption in France
According to INSEE's 2024 ICT Enterprises survey, 10% of French companies with 10 or more employees report using at least one AI technology. This figure rises to 33% among companies with 250 or more employees, confirming that large groups and mid-sized enterprises are moving faster than the rest of the business landscape.
This growth in usage is not always accompanied by an equivalent effort in terms of training and skills management. The 2025 Training and Employment Barometer (CSA/Centre Inffo) shows that 53% of workers now use AI tools in their professional activity, whilst 38% report having received no formal training on these technologies.
A Structural Gap Between Usage and Mastered Skills
AI is being integrated into daily practices faster than upskilling programmes can keep pace. This gap creates a paradoxical situation:
On one hand, organisations are experimenting with and deploying AI solutions, often driven by performance or innovation pressures.
On the other, the competencies needed to understand and exploit these uses remain only partially structured.
This imbalance is not inevitable. It signals an opportunity: to build a structured AI upskilling pathway aligned with existing practices, rather than allowing adoption to happen opportunistically.
Integrating AI Without a Method Creates Organisational Debt
When artificial intelligence is introduced without a clear framework, it generates organisational debt comparable to technical debt: rapid decisions that produce delayed effects that are difficult to correct.
Tool Proliferation and Fragmented Practices
The first source of organisational debt lies in the multiplication of AI solutions introduced over the course of various projects. Each team tests its own tools and explores new features without cross-functional coordination. This logic of permanent experimentation fragments practices and obscures the overall picture. It becomes difficult to identify who does what, against what benchmarks, and according to what rules of use.
Over time, this accumulation muddies understanding of the competencies actually being mobilised. Organisations have powerful tools at their disposal, but struggle to capitalise on the learning acquired.
Dependence on Experts and the Dilution of Accountability
Organisational debt also manifests as dependence on a small number of experts, frequently called upon to interpret results or secure usage practices. This concentration of knowledge creates an imbalance. Under these conditions, accountability becomes blurred. Who validates a recommendation generated by an automated system? Who adjudicates in the event of an error? AI embeds itself in processes without genuine organisational ownership.
AI Skills: Strengthening What Already Exists to Avoid Organisational Complexity
The most common mistake in AI integration projects is treating AI skills as a standalone bloc. This approach frequently leads to tools or roles being layered on top of one another, without clear articulation with the existing organisation.
Anchoring AI Skills in Existing Know-How
AI-related skills are not a standalone bloc: they build upon the professional expertise and competencies already mobilised within the organisation. Understanding a model or interpreting an automated recommendation rests first and foremost on a thorough knowledge of processes and operational constraints. Without this foundation, AI remains an isolated device, difficult to exploit sustainably.
Mastered AI skills integration therefore means starting from what already exists: skills frameworks, ways of working, decision-making processes.
The Skills Matrix as a Structuring Foundation
This is where structuring tools, such as the skills matrix, come into their own. It offers a cross-functional view of competencies already in place and facilitates the identification of foundations upon which to integrate AI without disruption, whilst adapting training pathways accordingly. By making visible the capabilities available within the organisation, it enables the construction of a coherent AI upskilling trajectory, firmly rooted in the operational realities of each team.
From Machine Learning to Generative AI: What Skills Are Required?
Behind the term "AI" lie very different approaches, each requiring different competencies. Clarifying these distinctions helps avoid a double pitfall: overestimating the technical dimension on one hand, or underestimating the human competencies required for the real-world use of AI on the other.
What AI Technologies Actually Encompass
Machine learning technologies rely on the analysis of large volumes of data to produce predictions or automate certain classifications. They are already widely used in predictive analytics and decision support. These systems most commonly rely on neural networks, capable of identifying complex patterns by progressively learning from input data.
Generative AI, by contrast, produces content or recommendations from existing data. The user formulates hypotheses, tests new features, and refines their requests based on the results obtained.
Within an HR function, a machine learning model can identify skills development trends from historical data. AI assistants can go further by acting as skills generators, proposing personalised upskilling pathways or suggesting training suited to a given profile.
In both cases, the critical competency remains the ability to interpret results, not to develop the model.
The Skills Expected Beyond the Technology
Whatever the technology used, the AI skills expected go well beyond tool proficiency. They rest on capabilities already present within the organisation, but called upon to evolve:
Understanding what an automated system does — and what it does not do;
Identifying biases or limitations in a generated output;
Placing a recommendation within its professional and regulatory context;
Deciding when to follow, adjust, or discard a suggestion produced by AI.
A manager using a recommendation from a generative AI system to anticipate skills needs is exercising a judgement competency that already exists — enriched by access to a larger volume of information. AI changes the scale and speed, not the accountability.
Managing an AI Skills Project Across an Organisation
At organisational level, the challenge is not simply to identify use cases, but to define who mobilises which skills, at what point, and with what level of accountability. Without this framework, AI takes hold opportunistically, shaped by local initiatives.
Clarifying Governance and Roles Without Creating New Layers
The first condition for effective management lies in governance. Embedding AI in practices means clarifying existing roles rather than systematically creating new ones. HR, learning, operational, and IT functions retain their respective responsibilities, but must share a common foundation of AI usage.
Managing AI skills also requires rethinking decision-making processes. AI makes it possible to produce analyses in real time or at scale, but these capabilities are only valuable if decision-makers have the competencies needed to exploit them.
Aligning AI Initiatives With a Legible Upskilling Trajectory
Managing an AI skills project means aligning project initiatives with a clearly legible upskilling trajectory. This requires integrating lessons learnt from ongoing AI projects into skills frameworks and training programmes, so that on-the-ground learning progressively enriches the organisation as a whole. AI upskilling is thus built incrementally, drawing on what already exists, without creating an artificial break with established practices.
