¿Qué es OpenAI LifeVal? El benchmark de tareas cotidianas
¿Qué es OpenAI LifeVal? El benchmark de tareas cotidianas OpenAI LifeVal es un benchmark de 169 tareas de la vida cotidiana, como matricular a un hijo en la escuela, gestionar el presupuesto de un hogar o tratar con una oficina gubernamental, que un agente de IA completa con uso del navegador, uso del ordenador y conectores. Una rúbrica puntúa cada intento, y la puntuación es la recompensa media…
OpenAI LifeVal is a benchmark created by OpenAI Research that measures how well an AI model performs in everyday tasks rather than programming tasks or exam questions. The benchmark consists of 169 tasks that require an AI agent to navigate the web, operate a computer, and use connectors. Each task is rated on a scale of 0 to 1, with the average reward across all attempts serving as the score.
GPT-6 Astra scored 80.6% using the xhigh effort level. OpenAI gathered data for LifeVal by analyzing real-life activities of people using the American Time Use Survey, the Multinational Time Use Study, and a survey of 3,995 demographically representative participants about the tasks they would delegate to an AI agent. The 169 tasks were then selected based on the intersection of these data sources.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
