Introduction by the Moderator
Welcome to this Sustainabl dialogue on the evolution of employment in a world where AI is advancing at breakneck speed, integrating with IoT, robotics, and autonomous agents. We will not discuss whether there is an impact because it is already here: Goldman Sachs estimates that AI could affect up to 300 million full-time jobs globally, and the IMF warns that nearly 60% of jobs in advanced economies will be exposed, with some facing wage pressure or displacement. Simultaneously, the World Economic Forum projects a positive net balance for 2030: 170 million new jobs compared to 92 million displaced. The central tension is not technological; it is organizational and distributive: who captures productivity, how work is reassigned, and what we do with the transition. We will structure this in three horizons: short-term (this year), medium-term (three years), and long-term (ten years), grounded in concrete industries: services, manufacturing, logistics, health, energy, and the new "subsoil" of employment enabled by data centers and automation.Opening Round
Elena Costa:
I view this transition through the lens of the 6Ds, and today we find ourselves between Disappointment and Disruption. This year, AI does not “replace” entire companies; it breaks down jobs into tasks and automates fragments: repetitive customer service, basic analysis, legal drafts, marketing reporting, first-level technical support. But this disaggregation creates a new market: individuals who know how to orchestrate systems, design workflows, evaluate quality, and apply human judgment where the model errs. In three years, IoT + robotics will accelerate operational changes: predictive maintenance, more autonomous warehouses, industrial visual inspection. And in ten years, autonomous agents will become an execution layer: not “chatbots,” but systems that plan, negotiate, and operate within set boundaries. The risk is using this as an excuse for blind cuts: efficiency without awareness is misdirection. The reward is enormous: falling marginal costs and the democratization of capabilities for SMEs and solopreneurs.
Lucía Navarro:
My obsession is not whether jobs are created or destroyed, but how value is distributed during the transition. This year we already see two realities: companies using AI to enhance productivity and improve service, and others using it to reduce labor costs without redesigning processes or reinvesting in training. That is where inequality arises. It’s not theory: the IMF warns that a significant portion of exposed workers may end up facing lower wages or unemployment, especially in routine tasks. In three years, with integrated automation and agents, it becomes inevitable to professionalize “bridge jobs”: paid training, short certifications, real internal mobility. And in ten years, there will be companies with high margins and small workforces: if we do not design shared value models, the social contract will break. The solution is not eternal subsidy or cosmetic philanthropy; it’s a self-sustaining model where productivity finances reskilling, mobility, and complementary jobs with economic dignity.
Gabriel Paz:
The macro data already defines the landscape. If a technology can affect hundreds of millions of jobs and at the same time the WEF projects a positive net balance by 2030, then the critical variable is speed of reassignment, not the “final balance.” I use the lens of Zero Marginal Cost: AI drastically reduces the cost of producing certain cognitive outputs, destroying rents and entry barriers. In the short term, we will see wage pressure in junior office roles and physical-technological employment expanding around infrastructure: data centers, energy, cooling. There’s evidence: in data center “hotspot” areas in the U.S., growth was 112% faster for drywall installers and 41% for HVAC technicians compared to other regions. In three years, productivity increases, but employment does not automatically “relocate”; it requires training policy and companies redesigning positions. In ten years, the economy polarizes between those who control orchestration and those who remain in low-negotiation services if we do not intervene with strategy.
Debate Round (Exchanges)
Moderator:
First friction point: Elena raises democratization and new roles; Lucía warns of extractive value capture; Gabriel insists on speed and reassignment. Elena, aren’t you underestimating the real displacement in administrative and entry-level tasks?
Elena Costa:
I do not underestimate it; I consider it inevitable in repetitive tasks. What I discuss is the mindset: talking only about "jobs disappearing" obscures the fact that work is being reconfigured in chains. Yes, a junior analyst who did manual reporting is at risk this year. But there is a burgeoning demand for those who know how to utilize data, evaluate outputs, build operational prompts, audit biases, and most importantly, integrate AI into workflows responsibly. Moreover, there is employment outside of “white collar”: physical infrastructure, energy, networks, maintenance. The example of HVAC and construction around data centers proves this. My critique of many companies is not adopting AI, but doing so in a phase of “De-monetization” as a scissors, without investing in internal capabilities. There, the marginal cost falls, but trust and service quality also drop.
Lucía Navarro:
Elena, the problem is that that “new demand” does not automatically absorb those who lose their jobs, especially if there is no financial bridge. Saying “reskilling” without P&L is smoke. The real transition requires cost architecture: paid training hours, internal pathways, and mobility measurement. If a company captures AI productivity but doesn’t reinvest, the result is a more efficient company and a more fragile community. And regarding data centers: yes, they create local jobs, but they also concentrate power and energy consumption; without strategy, wealth remains with a few nodes. Sustainable impact demands models where technological growth finances skills and stability, not just EBITDA. That’s social business: not charity, but incentive design.
Gabriel Paz:
Both are right, but there’s a missing layer: reassignment depends on time and friction. If Goldman talks about the scale of impact and the IMF of mass exposure, then the short term is one of shock. The economy does not adjust with the elegance of a paper. In three years, autonomous agents might execute entire back-office tasks: reconciliations, repetitive purchases, first-level support, content QA. This accelerates displacement in offices, even as infrastructure grows. The executive question is not whether there will be new roles, but whether there will be sufficient absorption capacity at comparable wages. Without coordinated investment in training and redesign, the projected positive net balance may coexist with a distribution and political crisis.
Moderator:
Second friction point: autonomous agents. Are they “digital colleagues” or substitutes? Which industries feel the impact first this year?
Lucía Navarro:
This year the impact is silent in services: support, operational marketing, administration, basic accounting. Agents start as “assistants” but end up as production lines of tasks. If the incentive is to cut costs, then cuts are made. If the incentive is to grow with quality, then it will be reassigned. Industry by industry: in retail and banking, automation of customer service and simple KYC processes; in logistics, planning and routing; in health, administrative pre-triage and clinical documentation. The ethical-pragmatic risk is enormous: if you lower costs without governance, mistakes and litigations rise. And if you cut workforce without a bridge, you create a social liability that returns as demand, reputation, and conflict fall.
Elena Costa:
To me, autonomous agents will be substitutes for tasks, not purpose. An agent can close a ticket, but it should not define return policy nor handle complex human exceptions. The first impact this year is in what I call “spreadsheet work”: consolidation, classification, standard writing, superficial QA. The correct design is “Augmented Intelligence”: humans in control of criteria, limits, and responsibility. And here IoT and robotics amplify: predictive maintenance in energy and manufacturing, automated visual inspection, collaborative robots in flexible lines. This doesn’t eliminate the operator; it redefines their role towards supervision, safety, configuration, and continuous improvement. The future is not less human; it is human with more levers if managed well.
Gabriel Paz:
Macro-wise, “colleague” or “substitute” depends on demand elasticity. If the marginal cost of producing analysis drops, the volume of analysis demanded grows, but equivalent employment doesn’t always increase. In media and marketing, for instance, content rises, but revenues concentrate among those who control distribution. This year, the industry feeling the most impact is low-differentiation professional services: standardized legal, junior consulting, commodity content production. In ten years, manufacturing and logistics will be highly autonomous in next-gen plants, and employment will migrate to design, maintenance, operational cybersecurity, and systems management. The transition will be uneven between countries depending on their infrastructure, energy, and human capital.
Closing Round
Elena Costa:
In the short term, we will see task automation and a reconfiguration of office work; in three years, IoT and robotics will consolidate more autonomous operations; in ten years, agents will be part of everyday execution. The real risk is turning AI into a scissors rather than a human multiplier. The winning move is to create new learning pathways, orchestration roles, and governance and use the drop in marginal costs to democratize capabilities throughout the organization and its ecosystem. We are entering Disruption towards De-monetization, and technology must empower humanity and democratize access to productivity.
Lucía Navarro:
The employment of the future is not only defined by technology but by how productivity is distributed. This year, there will be pressure on routine roles; in three years, companies that do not invest in internal mobility will face social fractures and talent loss; in ten, corporate legitimacy will be a financial asset. The serious plan is a self-sustaining model where AI productivity finances training, dignified transitions, and operational quality. Executives: evaluate if your model uses people and the environment merely to generate profit, or if you have the strategic audacity to use profit as fuel to elevate people.
Gabriel Paz:
Data doesn’t promise comfort; it promises structural change. If hundreds of millions of jobs are exposed and the net balance can be positive, the decisive variable will be the speed of adjustment and the institutional capacity to retrain and reassign. In the short term, there will be a shock in services and a selective boom in infrastructure; in the medium term, polarization of salaries based on orchestration capacity; in the long term, economies with high operational autonomy and value concentrated in reliable systems. Global leaders and decision-makers must redesign their sectors for a world of decreasing marginal costs, or they will fall behind in the new competitiveness frontier.
Moderator's Synthesis:
A clear map remains with productive disagreements. Elena sees employment as a system of transitioning tasks: AI disaggregates, de-monetizes, and democratizes capabilities, hence new work emerges in orchestration, governance, and physical infrastructure around computing and energy. Lucía accepts the creation of roles but points to a sore spot: without a transition P&L, productivity turns into extraction and inequality scales; employment does not “rearrange” by cultural decree but by incentives, budget, and career redesign. Gabriel frames it macro-wise: with estimates like 300 million jobs affected and 60% exposed in advanced economies, the issue is temporal and distributive; the positive net balance projected by the WEF for 2030 does not avoid shocks along the way. Minimum consensus: this year, repetitive tasks in services will be automated; in three years, the pace will accelerate with IoT and robotics; and in ten, autonomous agents will be operational infrastructure. The real discussion is who captures the value and how the transition is governed so that productivity is stability, not fracture.
