Rohan Paul· @rohanpaul_ai · X·· 2 小时前AI 评分67
AI 导读
OpenAI 发布 LifeVal 基准，用于测试 AI 模型能否替人完成日常事务。该基准依据数千名 ChatGPT 用户反馈构建，任务覆盖育儿、健康、教育、金融、购物和家庭事务，例如预约车辆年检、在预算内订购清真食品。Rohan Paul 称 GPT-6 Astra 在该基准上得分 80.6%，而不到 9 个月前 GPT-5.2 为 49.3%。
正文 · 原文
OpenAI has released LifeVal, a benchmark testing whether AI models can finish everyday errands for people.
GPT-6 Astra, scores 80.6% on everyday tasks, up from 49.3% for GPT-5.2 less than 9 months earlier.
LifeVal’s tasks span caregiving, health, education, finance, shopping and household admin, from booking a car inspection to ordering halal groceries within budget.
For AGI to benefit all of humanity, it should help people with the everyday tasks that matter most to them. So we built LifeVal, a new benchmark shaped by feedback from thousands of ChatGPT users on where they most want help from a personal assistant—from finding childcare to accessing government services to managing personal finances. Over the past year, we’ve used LifeVal to measure and improve how well our models handle these everyday tasks. https://openai.com/index/lifeval/在 X 查看被引用的帖子
来源：Rohan Paul · x.com