# sfeir-ia-emploi-risque-decrochage-2026-07-23

## Veille

In-depth opinion piece published on **sfeir.com** on July 23, 2026, signed by **SFEIR** (the firm's editorial voice). It is a **strategic commentary on Trésor-Éco note No. 391** from the DG Trésor (June 2026 — see [[dgtresor-ia-effets-emploi-2026-06-30]]), read through SFEIR's doctrine of « **amplifying AI rather than enduring it** ». The article praises Bercy's **cautious economist's tone** (mechanisms plus uncertainty rather than a prediction) and draws from it a **three-part thesis**: (1) **no measurable aggregate effect** at this stage (two offsetting forces — displacement vs. productivity — EU adoption ~20%); (2) a **single solid empirical signal, on juniors** (−16% employment among exposed 22-25 year-olds in the US); (3) a **long-term danger that shifts the question** — **competitive lag** (non-adoption), not job destruction. The analytical core SFEIR retains: **price elasticity** determines the employment effect (the **Jevons** paradox applied to code) → the argument is **structurally pro-employment for developers**. The article **dismantles the "AI layoffs" narrative** (4.5-6.2% of US layoff announcements, "labeling" at 59%) and points to the note's **blind spots** (the agentic scenario relegated to a footnote; diffusion speed not discussed; OpenAI/Anthropic having become sources for Bercy = an unflagged source bias). **SFEIR's operational translation** (for CIOs/CTOs): value migrates toward intent/architecture/control, training **augmented engineers** (**AI Champions** programs), and avoiding rushed adoption (**workslop**, technical debt) through **context engineering** and governance.

## Titre Article

IA et emploi : le vrai risque, c'est le décrochage

## Date

2026-07-23

## URL

https://www.sfeir.com/articles/ia-emploi-vrai-risque-decrochage/

## Keywords

AI and employment, competitive lag, non-adoption, Trésor-Éco 391, Bercy, DG Trésor, official doctrine, displacement effect, productivity effect, aggregate effect, AI adoption, price elasticity, Jevons paradox, code as commodity, pro-employment for developers, junior signal, youth employment, codified tasks, renewal of expertise, labeling, AI layoffs, Acemoglu Restrepo, Richmond OpenAI, Brynjolfsson, Arquié Coface, agentic scenario, creative destruction, diffusion speed, source bias, amplify rather than endure, augmented engineers, AI Champions, context engineering, workslop, technical debt, AI governance, POC to production, IT transformation, CTO, SFEIR, thought leadership

## Authors

**SFEIR** — ESN française « AI Only » (~850 ingénieurs, 8 agences France & Benelux). Voix éditoriale du cabinet (byline « SFEIR »). Positionnement de la maison sur la transformation IA des DSI ; ce texte prolonge la ligne éditoriale portée notamment par Didier Girard (cf. [[girard-sfeir-ai4it-vs-ai4business-budgets-2027-2026-06-24]]).

## Ton

**Profile**: a consulting-firm opinion piece (thought leadership), in a **strategic** register aimed at **CIOs/CTOs**, anchored to an institutional source. A stance of **critical reading**: SFEIR comments, praises, but also **corrects** Bercy's note (« what the note does not say loudly enough »).

**Style**: structured through thesis-subheadings (*Bercy states its doctrine → No aggregate effect → The only solid signal: juniors → The argument the debate misses → The "AI layoffs" narrative dismantled → The blind spots → The real fault line: competitive lag → SFEIR's perspective*). **Two movements**: first a **faithful, pedagogical restitution** of the note (mechanisms, figures, sources), then an **editorial turn** — the value of the analysis is reinterpreted into **decisions for the enterprise**. Epistemic honesty inherited from the note (« the note refrains from reading an established causality into it ») + **critical value-add** (the blind spots: agentic, speed, source bias). Signature phrases: *« amplify rather than fall behind »*, *« the Jevons paradox applied to code »*, *« when the bottom rung of the ladder disappears »*, *« turning adoption into a lever rather than a pile of POCs »*, *« adopting without method produces workslop and technical debt at industrial speed »*.

## Pense-betes

- **Thesis to remember**: SFEIR endorses and amplifies Bercy's note — **the real risk is not job destruction but competitive lag** (non-adoption). Bercy **shifts the burden of proof**: the danger lies less in adopting AI than in the **delay** in adopting it. It is a risk that is **competitive before it is social**.
- **The triptych to remember** (offered as a talking-points angle): *« no aggregate effect, a real junior signal, the risk is non-adoption »* — plus the **elasticity argument** on developers.
- **Jevons applied to code (the actionable core)**: a profession's fate is decided by the **price elasticity** of demand, not by exposure. Developers and graphic designers have an elasticity **> 1** → when AI lowers the cost of production, demand rises **more than proportionally**. Inelastic professions (firefighter, executive): unchanged volume. **Technical substitutability does not prejudge the employment effect.** → argument **structurally pro-employment for developers**. See [[wardley-llms-vibe-coding-developers-jevons-paradox-2026-03-27]].
- **SFEIR corollary**: if code becomes a **commodity**, value migrates toward **intent, architecture, control**, and software demand rises. The question shifts from *« how to produce cheaper code »* to *« what to build that we could not previously afford »*.
- **The only solid empirical signal = juniors**: −16% employment among exposed **22-25 year-olds** in the US (Brynjolfsson 2025, corroborated by Massenkoff-McCrory/Anthropic, Hosseini); France: −3.8 pts for **15-29 year-olds in IT** (Insee), 15-24 unemployment 19.1%→21.1%. **Mechanism**: AI automates the **codified tasks** of entry-level roles — *« those same tasks used to train tomorrow's seniors »*. The stake = **renewal of expertise**, not just entry into the workforce. → **organizational warning**: training **augmented engineers** is a deliberate choice (**AI Champions** logic).
- **"AI layoffs" narrative dismantled**: 4.5-6.2% of announced US layoffs in 2025, diluted within larger hiring flows; **labeling** (59% of US firms cite AI to justify freezes) = narrative cover for post-Covid corrections / investment reallocation. **Counterpoint**: OpenAI is doubling its headcount; Nadella quoted — **20-30% of Microsoft code already AI-generated, targeting 95% by 2030**.
- **The blind spots (SFEIR's critical value-add)**: 1. **Agentic scenario in a footnote**: it could **invalidate the "occasional assistant" framework** underpinning the "no aggregate effect" finding. Arquié: if agentic AI becomes widespread, **>40% of professions** would exceed 30% automatable tasks. Today's assessment describes a world of **assistance tools**, not **autonomous agents**. 2. **Creative destruction without discussing speed**: yes, 60% of today's jobs did not exist in 1940 (Autor), and Frey & Osborne 2013 (47% automatable) never materialized — **but the singularity of this wave is its diffusion speed**, left out of the framework. 3. **Source bias**: OpenAI and Anthropic now appear among Bercy's sources **on par with the NBER or the ECB** — the labs produce data on **their own impact** (unprecedented empirical richness + unflagged bias).
- **Exposure range** (reminder): 5%-60% depending on the study; Richmond (OpenAI) 18% high risk / 24% recomposition / 12% growth; Arquié (Coface) France 3.8% today → 16.3% within 2-5 years. **Exposure measures a potential, not actual employment.**
- **Public policy (drawn from the note)**: France holds an **intermediate** position (18% OECD adoption vs. 20% EU) but is **catching up faster** (+8 pts/year vs. +6). Programs: the **« Osez l'IA »** plan (15M trained by 2030, Académie de l'IA), **France 2030** "Skills and jobs of the future". Transition cost: slow reallocation cuts **−40%** of the benefit → **bridge jobs** + training.
- **SFEIR's operational translation (the punchline)**: *« standing still is the real risk, and rushed adoption is another »*. Adopting **without method** = **workslop** + technical debt at industrial speed. The trajectory depends on **context engineering**, clear **governance** of what is automatable, and **POC-to-production criteria**. *« Bercy names the danger of falling behind; our job is to turn adoption into a lever rather than a pile of POCs. »*
- **Meta / cross-reference**: this is the **"adoption/amplification" counterpart** to the AI4IT budget thesis [[girard-sfeir-ai4it-vs-ai4business-budgets-2027-2026-06-24]], and the **SFEIR commentary** on the official note [[dgtresor-ia-effets-emploi-2026-06-30]]. Same cluster as the announced disappearance of developers [[bfmtv-tech-co-business-ia-developpeurs-disparaissent-2026-05-05]].

## RésuméDe400mots

In this opinion piece published on sfeir.com (July 23, 2026), **SFEIR** comments on the **Trésor-Éco No. 391** note from the DG Trésor (June 2026) and anchors it to its own doctrine: *« amplifying AI rather than enduring it »*. The article praises Bercy's **cautious tone** — which lays out mechanisms and uncertainty rather than settling the matter — and draws from it a three-part thesis *« more reversed than it appears »*.

**No aggregate effect.** Within the Acemoglu-Restrepo framework, two forces oppose each other: the **displacement** effect (substitution) and the **productivity** effect (complementarity, lower costs, increased demand). They currently offset each other; studies identify no aggregate effect, for lack of hindsight and adoption (~20% of EU firms). Individual gains are nonetheless real (+14% in customer service, +26% among developers), but anxiety outpaces the data (62% of French people worried).

**The only solid signal: juniors.** −16% employment among exposed 22-25 year-olds in the US (Brynjolfsson 2025); in France, a contraction in youth employment in IT and rising unemployment among 15-24 year-olds (19.1%→21.1%) — without established causality. The mechanism: AI automates the **codified tasks** of entry-level positions, the ones that *« used to train tomorrow's seniors »* — hence a **renewal-of-expertise** issue.

**The argument the debate misses.** A profession's fate hinges on the **price elasticity** of demand, not exposure: developers and graphic designers (elasticity > 1) see demand grow as AI lowers their costs — the **Jevons paradox applied to code**. The argument is **structurally pro-employment for developers**. The article also **dismantles** the "AI layoffs" narrative (4.5-6.2% of US layoff announcements; **labeling** at 59%) and points to the note's **blind spots**: the **agentic** scenario relegated to a footnote (which would invalidate the "assistant" framework), **diffusion speed** left undiscussed, and **source bias** (OpenAI/Anthropic having become sources for Bercy).

**The real fault line: competitive lag.** Bercy shifts the burden of proof — the risk is **competitive** (falling behind in adoption), not social. Hence the programs (« Osez l'IA », France 2030).

**SFEIR's perspective**: for a CIO/CTO, this translates into decisions — value migrates toward intent/architecture/control; train **augmented engineers** (AI Champions); avoid rushed adoption (**workslop**, technical debt) through **context engineering**, governance, and POC-to-production criteria. *« Turning adoption into a lever rather than a pile of POCs. »*

## GrapheDeConnaissance

- SFEIR —publie→ IA et emploi : le vrai risque, c'est le décrochage (article) (DOCUMENT, 0.98)
- SFEIR (article) —référence→ Trésor-Éco n° 391 (DG Trésor) (DOCUMENT, 0.98)
- SFEIR —affirme_que→ le vrai risque de l'IA sur l'emploi est le décrochage compétitif (non-adoption), pas la destruction d'emplois (AFFIRMATION, 0.95)
- paradoxe de Jevons —s_applique_à→ la demande de code (élasticité > 1 chez les développeurs) (CONCEPT, 0.92)
- SFEIR —affirme_que→ l'argument d'élasticité-prix est structurellement pro-emploi pour les développeurs (AFFIRMATION, 0.9)
- élasticité-prix de la demande —permet→ de déterminer l'effet emploi d'un métier, davantage que son exposition à l'IA (AFFIRMATION, 0.9)
- IA générative —réduit→ l'emploi des juniors exposés (−16 % chez les 22-25 ans aux US) (MESURE, 0.85)
- automatisation des tâches codifiées —s_oppose_à→ le renouvellement des expertises (formation des futurs seniors) (CONCEPT, 0.85)
- SFEIR —affirme_que→ le scénario agentique, relégué en note de bas de page, pourrait invalider le cadre « assistant ponctuel » de la note (AFFIRMATION, 0.88)
- SFEIR —affirme_que→ OpenAI et Anthropic, devenus sources de Bercy, produisent la donnée sur leur propre impact (biais de source non signalé) (AFFIRMATION, 0.87)
- labellisation —observé_dans→ 59 % des entreprises US qui invoquent l'IA pour justifier des gels d'embauche (MESURE, 0.85)
- SFEIR —recommande→ former des ingénieurs augmentés (programmes AI Champions) face au signal juniors (AFFIRMATION, 0.9)
- SFEIR —recommande→ encadrer l'adoption par le context engineering et une gouvernance de l'automatisable (éviter le workslop) (AFFIRMATION, 0.9)
- adoption précipitée de l'IA —s_oppose_à→ qualité logicielle (produit du workslop et de la dette technique) (CONCEPT, 0.85)
- code —affirme_que→ quand le code devient une commodité, la valeur migre vers l'intention, l'architecture et le contrôle (AFFIRMATION, 0.88)

---
Canonical: https://www.thekb.eu/en/fiches/sfeir-ia-emploi-risque-decrochage-2026-07-23/