AI is replacing juniors: when short-term gains mortgage a company’s future

Immediate time savings, long-term skills loss

“Why hire an intern when ChatGPT can do the work in a few seconds, without needing supervision?” That’s a line I hear more and more in my digital transformation support engagements. Recently, the CEO of a shared-mobility scale-up told me he had stopped bringing in interns: generative AI now handles the research, preliminary analysis, and drafting tasks that traditionally make up a junior’s day-to-day work. An immediate productivity boost, sure. But at what cost for tomorrow?

This goes far beyond an isolated case. Recent announcements of massive AI investments to automate financial analysis work traditionally assigned to entry-level analysts confirm a deep, structural trend. Goldman Sachs and Morgan Stanley have publicly discussed cutting junior hiring by two thirds, arguing that generative AI can now take on a significant share of those tasks.

Even if this rationalization looks economically sound in the short term, it raises a fundamental question: how do you train tomorrow’s experts if you eliminate today’s learning roles? By prioritizing short-term efficiency, aren’t companies sawing off the branch that supports their future competitiveness?

A large-scale, well-documented phenomenon

The numbers speak for themselves. According to a Stanford University study published in August 2025, entry-level job postings in sectors vulnerable to automation fell by 13% [1]. This drop particularly affects two areas: software coding and customer service traditionally seen as gateways into the professional world.

Other studies confirm and amplify this finding. Korn Ferry reports that 37% of companies plan to replace entry-level roles with AI [2]. A survey by the British Standards Institution (BSI) goes further: 41% of executives surveyed say AI now enables headcount reductions, and 31% systematically consider an automation solution before launching a human recruitment process [3]. In large companies, half have already eliminated certain junior roles, compared with 30% in SMEs.

In France, APEC is also sounding the alarm: after a 19% decline in junior executive hires (less than one year of experience) in 2024, a further 16% drop is expected in 2025 [4]. IT, while still a major provider of executive jobs, saw junior recruitment plunge by 18% in 2024.

What’s emerging is a vicious circle: without entry roles, how can people gain the experience now required for nearly all positions? Companies seek experienced profiles capable of supervising AI tools, but refuse to train those who could become them. As a Stanford researcher notes: “The traditional springboard role offered by first-level jobs is disappearing.”

Risks for the company: beyond the obvious

The end of the talent pipeline

The first risk is mechanical: how will you have experienced seniors in 10 or 15 years if you stop training juniors today? An expert isn’t born an expert. They become one by going through every stage of learning: acquiring the theoretical basics, facing real-world situations, developing professional reflexes, peer transmission, and above all, accumulating experiences successes as well as failures.

This maturation takes time, typically at least a decade. By eliminating entry-level roles, companies dry up their own talent pipeline. Worse: they create total dependence on external hiring to fill positions of responsibility, a costly and risky strategy in a tight labor market.

The loss of company-specific know-how

Generative AI, as powerful as it may be, operates on “average” data. It produces statistically likely answers, formatted according to dominant practices absorbed during training. In other words, it excels at reproducing what already exists, but struggles to capture what makes an organization unique: its specific processes, its culture, the tradecraft accumulated over the years.

Companies hold an often underestimated intangible asset: tacit knowledge. This informal know-how, experts’ “tips and tricks,” reflexes in a specific situation, fine-grained understanding of a client or product context is neither documented nor truly documentable. It is passed on through observation, supervised practice, and daily exchange. Without junior profiles to receive and perpetuate this knowledge, it fades away as people leave.

According to several knowledge-management experts, a single senior departure can cause a loss of knowledge approaching one million euros when that expertise has not been transferred [5]. Scaled across an organization that no longer develops talent internally, the impact becomes considerable.

The risk of standardization and loss of differentiation

By relying heavily on the same generative AI tools drawing from broadly similar data corpora/companies converge toward increasingly homogeneous practices, content, and analyses. As an article by Crescera Solutions points out: “AI learns from masses of average data, so it generates average, consensus content that becomes more and more similar. Companies that overuse it lose their tone, their DNA, their uniqueness.” [6]

This invisible standardization is a direct threat to competitive differentiation. If all financial analysts use the same algorithms, if all marketing leaders rely on the same generative models, what will still set one company apart from another? Innovation capacity, diversity of thought, strategic agility these competitive advantages rest on the richness and uniqueness of human capital. A capital that mechanically impoverishes when training and knowledge transfer fade.

The generational break and its consequences

By closing the door to young talent, companies also create a dangerous generational rupture. Teams age without renewal, gradually losing touch with new ways of thinking, new expectations, and new habits. This disconnect can translate into a loss of market relevance, an inability to attract tomorrow’s customers, or growing difficulty innovating.

Moreover, this strategy contributes to a social fracture: young graduates end up in a professional dead end, deprived of the entry opportunities that traditionally opened access to higher-level roles. The societal and economic consequences of this mass exclusion could be major.

Reconciling AI and knowledge transfer: best practices

Given these observations, the challenge is not to reject AI, doing so would be as futile as it is ineffective, but to rethink its integration so it becomes a learning lever rather than a substitute for human training.

1. Reposition AI as an augmentation tool, not a replacement

The winning approach is to train “augmented juniors” rather than replace juniors with AI. Several studies show that an entry-level profile assisted by AI progresses faster and produces higher-quality work than a junior without AI and sometimes even than certain seniors who are less comfortable with these tools.

As one expert cited in multiple works reminds us: “The strongest is neither the human nor the machine, but the human who uses the machine.” This hybridization should be at the heart of training strategy: instead of eliminating entry roles, the goal is to transform them into “augmented learning roles,” where juniors learn both the craft and the mastery of its AI tools.

2. Redefine onboarding and training pathways

The most advanced companies adopt a methodical approach to integrating AI into training, prioritizing:

• Ambitious upskilling programs: Train everyone from executives to frontline teams to demystify AI and identify application opportunities in each domain of expertise.
• Concrete, progressive use cases: Structure learning across three levels, basic awareness, tool mastery, and role-based pathways integrating AI rather than forcing an abrupt revolution.
• Dedicated knowledge-transfer moments: Establish formal mechanisms for sharing know-how (workshops, senior–junior pairings, communities of practice) to ensure company-specific expertise keeps circulating.

3. Structure Knowledge Management

A robust knowledge-management approach becomes essential in this context. It aims to:

• Identify critical knowledge: Which know-how and skills are indispensable and cannot disappear? Which expertise is threatened in the short term (retirements, turnover)?
• Capture tacit and explicit knowledge: Document processes, but also organize the transfer of informal knowledge that cannot be codified.
• Facilitate access and sharing: Deploy collaborative platforms (internal wikis, knowledge bases) that centralize information while maintaining human practices of transmission.

Several companies report tangible results: a consulting firm halved the onboarding time of new consultants thanks to a shared knowledge base. An industrial group set up senior–junior pairings combined with filmed codification sessions, anticipating the retirement of 25% of its technical experts.

4. Treat knowledge transfer as a skill

Knowledge transfer must become a recognized, evaluated skill. In annual reviews, it is essential to ask about a person’s ability to transmit and to value contributors who invest in supporting juniors. This managerial recognition can take several forms: expert or “go-to” roles, participation in communities of practice, dedicated time in individual objectives.

5. Establish transparent dialogue about AI

AI integration should systematically be the subject of dialogue between employee representatives and management. Transparency on uses, impacts on roles, and support measures helps reduce fears (73% of French people worry about AI’s impact according to a Sia Partners study [7]) and co-build responsible adoption.

The most effective companies take a methodical approach involving close collaboration between top management, business teams, IT, and HR. They prioritize strategic areas where AI augments human capabilities rather than simply replacing them.

6. Maintain adapted entry-level roles

Rather than removing junior roles, the idea is to evolve them. The most repetitive tasks can indeed be automated, but entry-level profiles remain essential to:

• Learn to steer and control AI (verification, adjustment, improving outputs)
• Develop the critical judgment to know when to trust AI and when not to
• Acquire core professional fundamentals that enable progression to more complex responsibilities
• Bring a fresh perspective and challenge established practices

Conclusion: the urgency of a long-term vision

Artificial intelligence is neither a threat nor a miracle solution. It is a powerful tool that depending on how we integrate it can either increase our collective capabilities or impoverish them for the long run.

Today’s choice for CEOs, HR leaders, and digital transformation managers is clear: prioritize immediate productivity gains at the risk of mortgaging the future, or invest in a balanced approach that reconciles technological innovation with human-capital development.

The companies that maintain training efforts, structure knowledge transfer, and position AI as an amplifier rather than a substitute for human skills will be the ones that preserve their innovation capacity, strategic agility, and competitive differentiation.

Because beyond algorithms and efficiency gains, it is always people their expertise, creativity, and ability to think differently who create an organization’s value. Let’s not forget that in the race to automation.

Références

[1] Stanford University (November 2025). Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence, Erik Brynjolfsson, Bharat Chandar and Ruyu Chen. URL: https://digitaleconomy.stanford.edu/publications/canaries-in-the-coal-mine/

[2] Korn Ferry (October 2024). Rapport sur les tendances de recrutement et l'IA. Cited in LeMagIT, “Remplacer les juniors par l'IA, pas une bonne idée”, 10 November 2025. URL: https://www.lemagit.fr/actualites/366634321/Remplacer-les-juniors-par-lIA-pas-une-bonne-idee

[3] British Standards Institution (BSI) (October 2025). Étude sur l'impact de l'IA sur le recrutement. Cited in EconomieMatin, “40% des entreprises vont passer à l'IA plutôt que d'embaucher des jeunes”, 10 October 2025. URL: https://www.economiematin.fr/impact-intelligence-artificielle-emploi-jeunes-diplomes-debut-carriere

[4] APEC (2024–2025). Baromètre de l'emploi cadre. Cited in Clockia, “Premier emploi tech et IA : la double peine des jeunes diplômés ?”. URL: https://clockia.avignon-et-moi.fr/ia/3054-premier-emploi-tech-et-ia-la-double-peine-des-jeunes-diplomes.html

[5] Smart Tribune (January 2025). “Qu'est-ce que le Knowledge Management ? Définition et enjeux 2025”. URL: https://blog.smart-tribune.com/fr/knowledge-management-definition-enjeux

[6] Crescera Solutions (July 2025). “Intelligence Artificielle en entreprise : entre promesses, standardisation et enjeux de fond”. URL: https://crescerasolutions.com/intelligence-artificielle-en-entreprise-entre-promesses-standardisation-et-enjeux-de-fond/

[7] Sia Partners (June 2025). “IA vs RH : Chartes éthiques, des boussoles indispensables pour naviguer en eaux troubles”. Cited in Lamy Liaisons. URL: https://www.lamy-liaisons.fr/eclaireurs-du-droit/ia-vs-rh-par-sia-chartes-ethiques-des-boussoles-indispensables-pour-naviguer-en-eaux-troubles/