{"id":18681,"date":"2026-05-11T07:36:26","date_gmt":"2026-05-11T10:36:26","guid":{"rendered":"https:\/\/www.fie.undef.edu.ar\/ceptm\/?p=18681"},"modified":"2026-05-11T07:36:26","modified_gmt":"2026-05-11T10:36:26","slug":"ai-fight-club-lockheed-martin-revoluciona-el-entrenamiento-autonomo-simulando-114-anos-de-combate-aereo-en-un-entorno-sintetico","status":"publish","type":"post","link":"https:\/\/www.fie.undef.edu.ar\/ceptm\/?p=18681","title":{"rendered":"AI Fight Club, Lockheed Martin revoluciona el entrenamiento aut\u00f3nomo simulando 114 a\u00f1os de combate a\u00e9reo en un entorno sint\u00e9tico"},"content":{"rendered":"

Lockheed Martin ha marcado un hito en el desarrollo de sistemas aut\u00f3nomos mediante su iniciativa “AI Fight Club”, un entorno sint\u00e9tico donde se simularon 114 a\u00f1os de combate a\u00e9reo de alta intensidad. Utilizando el ecosistema digital “Cogniverse”, el ejercicio enfrent\u00f3 a diversos agentes de inteligencia artificial en escenarios t\u00e1cticos de 4 contra 4, evaluando su capacidad de toma de decisiones bajo presi\u00f3n extrema. Este enfoque permiti\u00f3 procesar datos de 18 millones de aeronaves derribadas virtualmente, logrando en semanas lo que requerir\u00eda d\u00e9cadas de pruebas de vuelo reales. Con este avance, la compa\u00f1\u00eda busca estandarizar un modelo de “simulaci\u00f3n primero<\/em>” para perfeccionar la autonom\u00eda de sistemas militares en m\u00faltiples dominios, desde el espacio hasta el \u00e1mbito mar\u00edtimo. Este evento no solo acelera la madurez t\u00e9cnica de la IA, sino que redefine c\u00f3mo se preparar\u00e1n las fuerzas para las futuras guerras en entornos altamente disputados.<\/p>\n

\n

Lockheed Martin\u00a0used<\/a>\u00a0a large-scale synthetic combat environment to simulate the equivalent of 114 years of AI warfare testing, as part of efforts to accelerate autonomous combat system development.<\/p>\n

The simulations modeled more than $540 trillion in operational costs and the virtual loss of 18 million aircraft during high-intensity aerial warfare scenarios.<\/p>\n

The figures come from a high-fidelity simulation framework designed to stress-test autonomous decision-making in complex air combat scenarios rather than represent real-world expenditure or losses.<\/p>\n

The environment is intended to provide a scalable way to evaluate how AI systems behave under sustained operational pressure.<\/p>\n

The\u00a0exercise<\/a>, branded the \u201cAI Fight Club,\u201d was conducted by Lockheed Martin in collaboration with Ansys Government Initiatives and ATG, Inc., bringing together AI agents controlling simulated fourth-generation fighter aircraft in 4-vs-4 aerial engagements.<\/p>\n

The event evaluated how autonomous agents make operational decisions, execute missions, and reduce losses in contested environments.<\/p>\n

Observers monitored five AI agent teams competing in real-time synthetic air combat while collecting operational data on their performance.<\/p>\n

\"\"
A simulated air combat scenario is displayed during Lockheed Martin\u2019s \u201cAI Fight Club\u201d autonomous warfare testing event. Photo: Lockheed Martin<\/figcaption><\/figure>\n

The scenarios were run inside Cogniverse, Lockheed\u2019s synthetic testing environment designed to replicate the complexity and dynamics of tactical air combat.<\/p>\n

\u201cThis exercise shows how combining trusted mission-level simulation with AI-driven insight provides the ability to explore scenarios, anticipate outcomes and validate operational plans before forces are ever deployed,\u201d said vice president and general manager of Ansys Government Initiatives\u00a0Mike Pedaci<\/strong>.<\/p>\n

AI Fight Club<\/strong><\/p>\n

The\u00a0AI Fight Club<\/a>\u00a0initiative, first introduced in 2025, is positioned as a defense AI proving ground where companies can test autonomous systems against realistic operational scenarios aligned with Department of Defense standards.<\/p>\n

Lockheed said the approach allows AI systems to be evaluated and refined without the cost, risk, and logistical constraints of live exercises.<\/p>\n

\u201cThe AI Fight Club event was a glimpse at how defense autonomy development should work. Large-scale, simulation-first and collaborative,\u201d said ATG director of operations\u00a0Scott MacDonald<\/strong>.<\/p>\n

Future iterations are expected to expand into additional platforms and operational domains, including homeland defense, maritime operations, and space-based surveillance and reconnaissance scenarios, as the initiative evolves toward joint all-domain operations.<\/p>\n