A humanism of gradients, capable of grappling with complexity without fear
En una sala de conferencias de Barcelona, una academia española ha planteado esta semana una pregunta que trasciende la tecnología: ¿puede la inteligencia artificial aprender a razonar como los seres humanos realmente deciden, en gradientes y no en certezas? La Real Corporación de Economistas de España propone fusionar la lógica difusa con la IA para construir lo que llama 'razonamiento artificial humanístico', un proyecto intelectual de cuatro décadas que desafía los fundamentos mecanicistas de la economía clásica. En un momento en que la UE regula la transparencia algorítmica y el Tribunal Supremo español la reconoce como valor democrático, esta propuesta sitúa la ambigüedad y la subjetividad no como defectos del pensamiento humano, sino como su característica más esencial.
- La inteligencia artificial, tal como se desarrolla hoy, opera sobre certezas binarias que no reflejan cómo los seres humanos realmente toman decisiones, lo que genera una brecha creciente entre las herramientas y la realidad.
- La RACEF lleva cuatro décadas construyendo una matemática de la incertidumbre que ahora choca con la urgencia global de regular la IA antes de que sus riesgos se vuelvan inmanejables.
- El concepto de 'simultaneidad gradual', introducido en 1986, dinamitó el principio lógico de tercero excluido y abrió paso a modelos económicos capaces de formalizar la subjetividad y la ambigüedad.
- La academia publica algoritmos gratuitos en su web y se posiciona como referente en transparencia algorítmica, anticipándose al debate europeo sobre los límites éticos y operativos de la IA.
- El debate se extiende desde la innovación regional hasta la energía nuclear, con un hilo conductor común: las viejas certezas ya no bastan para gobernar un mundo genuinamente complejo.
En Barcelona, la Real Corporación de Economistas de España reunió esta semana a su liderazgo para defender una tesis incómoda: la inteligencia artificial lleva décadas resolviendo el problema equivocado. Su propuesta, bautizada como 'razonamiento artificial humanístico', no parte de las certezas que han gobernado el pensamiento económico, sino de la realidad gradiente y desordenada en que los seres humanos realmente deciden.
El argumento central lo formuló Vicente Liern: la IA necesita lógica difusa, una matemática que no trata la verdad y la falsedad como estados binarios sino como cuestiones de grado. Joan Francesc Pont fue más lejos: el humanismo del siglo XXI no puede ser un humanismo de certezas, sino de gradientes, capaz de habitar la incertidumbre sin rendirse y de preservar la dignidad humana sin condiciones.
Este planteamiento tiene raíces profundas. En 1986, el presidente de la academia, Jaime Gil Aluja, introdujo en Buenos Aires el principio de simultaneidad gradual: una afirmación puede ser verdadera y falsa a la vez, siempre que se asignen grados a cada condición. Ese principio demolió la ley del tercero excluido, sobre la que descansaba toda la ciencia económica. En 1999, Gil Aluja fue más allá y desarrolló una matemática no numérica para la incertidumbre, capaz de formalizar la subjetividad. Su ejemplo es revelador: el precio de un objeto es el mismo para todos, pero su valor varía según quien lo recibe. La matemática clásica no puede capturar esa diferencia. La lógica difusa, sí.
La academia se presenta ahora como guardiana de la transparencia algorítmica, un principio que la UE ha comenzado a regular y que el Tribunal Supremo español reconoció recientemente como valor democrático. La RACEF ha publicado una docena de algoritmos de uso libre, construidos sobre principios de relación, asignación, agrupación y ordenación, con la intención de ofrecer respuestas científicas a dilemas éticos antes de que el debate global sobre la IA se vuelva urgente.
Las presentaciones abarcaron desde la competitividad regional del País Vasco hasta la defensa pragmática de la energía nuclear en Europa. El hilo conductor fue siempre el mismo: el mundo es complejo, la incertidumbre es real, y las herramientas con las que tomamos decisiones deben reconocer ambas cosas. Las viejas certezas ya no sirven.
In a Barcelona conference room this week, a Spanish academy made a quiet argument that artificial intelligence has been solving the wrong problem. The Real Corporación de Economistas de España gathered its leadership to discuss what it calls "humanistic artificial reasoning"—a framework built not on the certainties that have long governed economic thought, but on the messy, gradient-filled reality of how humans actually decide.
The conversation began in Crete, where academics from the RACEF presented alongside colleagues from the Technical University of Crete and other European institutions. Vicente Liern, one of the academy's senior members, posed the central challenge: artificial intelligence, for all its power, lacks something essential. It needs fuzzy logic—a mathematical approach that treats truth and falsehood not as binary states but as matters of degree. This fusion, he argued, points toward a genuinely humanistic form of machine reasoning.
Joan Francesc Pont pushed the idea further. "Twenty-first century humanism cannot be a humanism of certainties," he said. Instead, it must be a humanism of gradients—one capable of grappling with complexity without fear, of dwelling in uncertainty without surrender, of preserving human dignity without conditions. The distinction matters. For decades, economic models have treated the world as a machine governed by mechanical laws. Fuzzy logic and the mathematics that follow from it acknowledge that the world resists such simplification.
This represents a 40-year intellectual arc. In 1986, Jaime Gil Aluja, the academy's president, introduced what he called the principle of gradual simultaneity at a Buenos Aires conference. The idea was radical: a statement could be both true and false at once, provided you assigned degrees or levels to truth and falsehood. That single principle dismantled the law of excluded middle—the logical foundation on which all economic science had rested. Once entropy, chaos theory, uncertainty, and fuzzy sets entered the conversation, the door opened to something new: economic humanism.
Gil Aluja developed this further in 1999, creating a non-numerical mathematics for uncertainty. It allows economists and policymakers to formalize subjectivity and ambiguity in their decisions, moving away from the mechanistic models that had dominated the field. He offered a simple example: the price of an object is the same for everyone, yet its value differs from person to person. The price reflects the sum of materials, labor, distribution—but the utility, the actual worth to the buyer, depends on acceptance, on subjective judgment. Two people pay the same amount and receive different value. Classical mathematics cannot capture this. Fuzzy logic can.
The academy is now positioning itself as a guardian of algorithmic transparency—a principle the European Union has begun to regulate and that Spain's Supreme Court recently recognized as a democratic value. The RACEF has published a dozen algorithms on its website, free to use, built on principles of relationship, assignment, grouping, and ordering. These are meant to offer scientific answers to ethical and operational challenges, and to anticipate the global debate about AI's risks and limits before that debate becomes urgent.
Arturo Rodríguez presented research on regional innovation and competitiveness, comparing the Basque Country to Dutch regions. Ricardo Hernández argued for maintaining nuclear energy in Europe, citing Spain's Almaraz plant as one of the world's best—not from ideology, he said, but from pragmatism. The point was consistent across the presentations: the world is complex, uncertainty is real, and the tools we use to make decisions must acknowledge both. The old certainties no longer serve.
Citações Notáveis
AI's power must be combined with fuzzy logic to achieve humanistic artificial reasoning that handles complexity and uncertainty— Vicente Liern, RACEF academic
Twenty-first century humanism must be a humanism of gradients, capable of grappling with complexity without fear and dwelling in uncertainty without surrender— Joan Francesc Pont, RACEF academic
A Conversa do Hearth Outra perspectiva sobre a história
Why does an academy of economists care about fuzzy logic? Isn't that a mathematics problem, not an economics one?
Because economics has been pretending the world is simpler than it is. We've built models on the assumption that things are either true or false, that prices reflect objective value. Fuzzy logic says: no. A price is the same for everyone, but what it's worth to you depends on whether you want it. That's not a flaw in the data—that's the data.
So this is about admitting that human judgment matters in economic decisions?
Exactly. And not just admitting it—building mathematics that can actually handle it. For 40 years, we've been trying to force subjectivity into objective equations. It doesn't work. Fuzzy logic lets you formalize the gradients, the in-betweens, the uncertainties.
The academy is also talking about algorithmic transparency. How does that connect?
If we're going to use AI to make decisions that affect people's lives, those decisions need to be explainable. Not just accurate—transparent. The algorithms the academy published are designed to show their reasoning, not hide it. That's a democratic principle, not just a technical one.
Is this a European thing, or is Spain leading something here?
Spain is leading. The Supreme Court recognized transparency as a democratic value. The EU is regulating it. But the RACEF is ahead of both, offering tools and frameworks that other institutions can actually use. They're not waiting for the debate to settle—they're shaping it.