OpenAI has introduced LifeVal, a benchmark measuring how well AI helps with everyday tasks like finding childcare and managing finances. The goal is to build a personal assistant that actually gets things done.
To identify the most valuable tasks, OpenAI analyzed the American Time Use Survey and a survey of 3,995 respondents, focusing on caregiving, health, education, and government services.
Current models score highly on these practical workflows. For instance, they can help enroll children in kindergarten with 93.1% accuracy, organize flood volunteers at 98.0%, find a new home with 100% success, and set up remote game watching at 96.3%.
Over the past year, OpenAI has trained its models to be faster and more capable in browser and computer use, expanding the types of everyday tasks ChatGPT can successfully complete.
Source: OpenAI Blog · Summarized by HeadlinesBriefing