AI in Business: Skills Gap Threatens Innovation

The Massive Adoption of AI: An Untapped Potential

The integration of artificial intelligence into businesses has become an unavoidable reality. However, this widespread adoption does not necessarily guarantee a successful digital transformation. Hamza Bouanani, AI Practice Manager at MARGO and Lead Data Scientist at BNP Risk, highlights a major issue: the lack of structured training hinders the optimal use of AI tools. Although companies are investing heavily in software licenses, Bouanani observes a chronic underutilization of these technologies. Inaction regarding training now threatens the very survival of organizations.

“The transition to structured production is often an illusion,” says Bouanani. Indeed, despite the availability of tools, their use remains superficial for the majority of employees. Advanced features, particularly in coding, are mastered only by a technical elite. Among developers, only 10% fully utilize advanced coding agents, while half completely ignore AI in their daily workflow.

The Illusion of Equipment: Dormant Tools

The current situation reveals a paradox: while companies have crossed the threshold of equipping themselves with AI, they remain stuck in a state of underutilization. Office suites like Google Workspace AI offer immense creative possibilities, yet employees continue to design their presentations manually. This lack of methodological reflexes leads to wasted time and keeps AI at the level of mere technical demonstration.

“Even among regular users, the vast majority only exploit 5 to 10% of the tool's overall potential,” notes Bouanani. The absence of converting tools into structured business processes means that technological investment does not yield significant returns. Chronic underutilization then becomes an invisible cost for the organization.

The Four Pillars of Serious Training

A simple demonstration of ChatGPT is not enough for skill enhancement. The intuitive interface of generative AI creates an illusion of immediate mastery, but this ease of access masks critical methodological gaps. Bouanani identifies four major gaps among unsupported users.

The first obstacle is dialogue with the machine. Many employees use AI as a search engine, typing keywords without providing specific context. The results obtained are then flat and generic, which discourages the user and increases the rate of premature abandonment.

Data security is the second essential pillar. In urgency, teams copy confidential codes to public models, thereby exposing the company's intellectual property to massive leaks. Without a framework, the employee prioritizes efficiency over security.

Validating results also requires specific learning. AI crafts its responses with a narrative confidence that can be misleading, and employees often accept these outputs as absolute truths. It is crucial to learn how to contextualize the machine, a survival skill in the professional world.

Finally, the ability to chain complex tasks is lacking. AI should not only correct occasional spelling mistakes but also analyze reports, extract data, and generate action plans. This integrated workflow demands a comprehensive view of the business.

The End of Non-Augmented Profiles

Data literacy is becoming the new recruitment standard. Bouanani compares this evolution to the arrival of Excel, where knowing how to use a spreadsheet distinguished high-performing candidates. In 24 to 36 months, “data literacy” will be a disqualifying prerequisite for any position.

Companies will no longer be able to manage profiles incapable of interacting with algorithms. Marketing and customer relations are already experiencing this shockwave. A modern marketer must segment their databases using AI. Finance and operations are following this trajectory, moving away from the static logic of traditional pivot tables to querying models in natural language.

Domain Expertise Saves Technical Execution

The cliché opposing resistant seniors to agile juniors is misleading. The reality shows a more nuanced dynamic. The technical ease of younger profiles often conceals a lack of critical perspective. They manipulate the interface without always understanding the underlying business stakes.

Juniors do not fear the blank page when facing AI, but they validate incorrect answers due to a lack of industry knowledge. A subtle reasoning error easily escapes their vigilance. Experienced profiles provide the necessary compass for the generative tool. Despite initial apprehension, seniors derive greater value from AI. Their ability to delegate tasks precisely significantly enriches the prompts. The senior uses artificial intelligence as a debate partner. They immediately spot an anomaly in a margin calculation or a text. The marriage of both strengths creates true added value.

Internal Fracture, Retention, and Human Risk of Cognitive Asymmetry

The real threat lies in creating a two-speed workforce. The lack of widespread training dangerously segments teams. Some employees become “super-employees” augmented by technology, while others remain stuck in time-consuming and exhausting manual methods.

This asymmetry causes friction during daily collaboration. Trained employees expect a velocity that others cannot match, generating a profound sense of inequity within departments. The untrained become involuntary bottlenecks for production. If the company does not democratize access to AI, it creates a segregation between “super-employees” and collaborators relegated to the status of mere executors.

We also end up with deliverables of heterogeneous and uncontrollable quality. Some employees sometimes use AI in secret to save time, producing documents filled with platitudes or factual errors. The company then loses control over its brand signature.

Finally, the risk affects the retention of top technical talent. A trained expert will not stay in a technologically obsolete environment. They will join the competition to fully exploit their new augmented skills. Conversely, the untrained employee sinks into anxiety over their own obsolescence. The trained and seasoned talent in AI will develop significant frustration if they operate in an ecosystem that does not match their pace.