During the sixth edition of the webinar “North America — Artificial Intelligence” organized by the Foreign Trade Advisors, the tone was set from the opening. “AI will transform work much faster than expected, without us yet knowing whether the net employment balance will be positive or negative.“
The subject is no longer theoretical. It is strategic. It engages corporate competitiveness, organizational structuring, talent trajectories and, more broadly, the economic balance of nations.
Main guest of this session, Vanessa Lyon, Senior Partner at BCG based in New York, offered a lucid, documented and nuanced reading of AI’s impact on employment. Her intervention poses a central question: are we facing a classic productivity shock, comparable to the great industrial revolutions, or a structural transformation of the very sources of competitive advantage?
Several Waves of AI, Several Impacts
First methodological point: move beyond conceptual confusion.
“There’s not just ChatGPT when we talk about AI, there are many families of AI.“
Vanessa Lyon distinguishes four major waves:
- Predictive AI, derived from traditional machine learning, applied to pricing, demand or optimization.
- Generative AI, based on Large Language Models, capable of producing structured content.
- Agentic AI, combining the previous ones, capable of operating entire processes autonomously.
- Physical AI, still emerging, oriented towards advanced automation and the physical world.
This breakdown is crucial. Because the impact on employment varies according to technological maturity and the depth of integration into business processes.
AI Maturity and Performance: A Strong Correlation
BCG’s annual studies highlight continuous progression towards what is called “AI scaling,” that is, the shift from isolated experimentation to genuine industrialization of use cases. As Vanessa Lyon puts it, “the more you ride a wave, the more capable you are of riding the next wave.” Companies most advanced in this trajectory show significant positive correlations in terms of revenue growth, Total Shareholder Return, return on invested capital, and patent creation. The explanation lies less in technological performance alone than in organizational structuring. Mature actors are those who have established robust data governance and know how to integrate information at the heart of their decision-making processes. This accumulation of learning and integration generates a self-reinforcing competitive advantage.
Employment: Destruction, Displacement or Recomposition?
Public debate oscillates between alarmism and technological optimism. The data presented calls for analytical caution.
In the case of software developers, payroll analyses show a differentiated phenomenon: decline in hiring for junior profiles and maintenance, even growth, for senior profiles.
“Young graduates have less place today in software development.“
However, Vanessa Lyon introduces several nuances. The observed decline is partly part of a post-Covid adjustment, after a period of sometimes excessive hiring, and the observed productivity gains are not exclusively attributable to AI. In other words, the ongoing transformation does not follow a linear or homogeneous trajectory; it varies according to generations, industry sectors and functions within organizations.
The Team Pyramid Is Being Reconfigured
In the software development cycle, AI reduces the need for intermediate developers. But it increases requirements for: product definition, cybersecurity, architecture, system orchestration.
“There’s a need for fewer developers… but there are challenges in terms of security and more complex ecosystems.“
We are not facing a simple reduction in headcount. We are observing a mutation of the skills structure.
Productivity and Performance Redistribution
A study discussed during the webinar highlights a significant shift in the distribution of individual performances after AI integration. The overall average increases, while the relative hierarchy between profiles is reconfigured. On repetitive tasks, the least performing employees benefit from substantial support; conversely, on complex tasks, the most experienced profiles see their capacity amplified by a genuine multiplier effect. “The cards are reshuffled,” summarizes Vanessa Lyon. This dynamic requires a profound rethinking of talent evaluation models: far from standardizing performances, AI modifies their distribution and redefines value criteria.
The Critical Challenge: Moving from Tool to Redesign
One of the strongest messages of the intervention concerns integration.
“If we don’t integrate technology into processes, it will pile up.“
Vanessa Lyon mentions the risk of a “Lotus Notes effect”: accumulation of non-integrated tools creating technological wastelands. Two models oppose each other: AI as support: a new tool, like Excel and AI at the center: process redesign around agents.
“Think about how I put AI first… and how I organize my human resources around it.“
This conceptual shift is major. It moves the reflection from incremental optimization to organizational reconfiguration.
Middle Management: Tension Zone
The traditional pyramidal organizational model is now profoundly questioned. Vanessa Lyon emphasizes that “middle managers… will probably be jobs that are greatly disrupted.” In an AI-first logic, the structure evolves towards fewer hierarchical layers, more orchestration, increased transversality and the integration of AI agents acting as capacity multipliers. This transformation is not just an operational adjustment; it raises a genuine internal political question: will the current transmission belts accept redefining their role and scope of influence within the organization?
Investing in Human Capital: Condition for Scaling
Companies that fundamentally rethink their workflows invest more in training, operational support, rigorous tracking of value creation and role clarification; the result is significantly freed-up time, refocusing on strategic tasks, increased employee satisfaction and tangible improvement in decision quality“We must force a structured approach.“
AI without human transformation is ineffective. Technology alone does not produce value.
Junior Skills: A Structural Challenge
A central question emerged from the exchanges: how do we train future experts if entry-level positions become scarce? Vanessa Lyon acknowledges without detour that this is “a real question today.” Some suggest that the technological leverage effect could accelerate professional trajectories, allowing young talents to progress faster. But she immediately alerts to a decisive point: “the question of critical thinking remains extremely important.” She illustrates this with a very concrete observation: “when juniors make slides with ChatGPT, you see it immediately. It looks good, but it means nothing.” In other words, far from lightening intellectual requirements, AI imposes additional rigor in analysis, verification and thought structuring.
Data and Information Sovereignty
Another major strategic challenge: data urbanization. Vanessa Lyon insists that “there’s a very big data urbanization issue.” This problem opens a series of structuring questions: what data can be shared without risk, what information must remain internal, and how to preserve an information advantage in a competitive environment? In a context where AI systems are capable of massively aggregating and cross-referencing heterogeneous sources, the boundary between public information and sensitive data tends to blur, requiring much more rigorous governance.
AI and National Competitiveness
The historical parallel with the steam engine is invoked to illuminate contemporary issues. Guillaume Bouvard recalls that China allegedly “missed” the industrial revolution on an institutional level, failing to adapt its structures to technological changes. In this perspective, Vanessa Lyon observes that “there is really this hope that AI will allow the United States to regain an edge.” Artificial intelligence is thus considered as a major macroeconomic lever. It is likely to offset the demographic shock, sustainably revive productivity gains and open the way to new strategic industries, from space to suborbital labs to biotechnology.
Should We Fear 20% Unemployment?
The question is asked bluntly. The answer aims to be measured. The great technological revolutions of the past have displaced jobs more than they have eliminated them. Agricultural or domestic automation, for example, has not led to the establishment of generalized universal income. Transitions have been socially complex, sometimes painful, but they have been historically absorbed by the economy. As Vanessa Lyon recalls, “there have been phases of technological progress with always more work.” A major unknown remains: the current speed of transformation. “We’re at the very beginning… the early stages,” she emphasizes, leaving open the question of the pace and magnitude of adjustments to come.
A Two-Speed Economy?
Patricia synthesizes the exchanges by formulating a clear observation: “We have the impression that there is a two-speed economy.” Large companies, already structured and equipped with significant investment capacities, progress more rapidly in AI integration. Conversely, SMEs and less equipped organizations risk progressive disconnection. The risk is therefore not only technological; it is above all organizational and strategic.
In closing, Vanessa Lyon shares a personal conviction: “I strongly believe in technological progress.” She mentions advances in medical research, nuclear energy and space. She compares the current situation of AI to the early days of aviation: “A hundred years ago, we were still on flying wings.” We are going through an intermediate phase, where organizational models adapted to artificial intelligence are not yet stabilized. “Productivity requires transformations of the operating model that we still have difficulty visualizing,” she reminds us, emphasizing that the real challenge lies less in technology than in its structural implementation.
Ultimately, AI does not mechanically destroy employment; it redefines the structure of organizations, the hierarchy of skills, the sources of competitive advantage, data governance and, more broadly, the very nature of leadership. The central message is strategic: the challenge is not simply to adopt AI, but to redesign the organization around it. For leaders, the question is no longer “should we go?” but rather: how to structure the transformation so that AI becomes a lever for innovation rather than a fragmentation factor? We are, in Vanessa Lyon’s words, “at a crossroads.” What follows will depend less on algorithms than on our collective ability to rethink the economic and social model that hosts them.
