OpenAI Introduces LifeVal, a Benchmark for Everyday Personal Assistant Tasks

Key Info

OpenAI announced LifeVal, a new benchmark built from feedback from thousands of ChatGPT users about where they most want help from a personal assistant, and says it has used the benchmark over the past year to measure and improve its models.

Highlights

• LifeVal targets everyday tasks users care about most, including finding childcare, accessing government services, and managing personal finances.
• The benchmark was shaped by feedback from thousands of ChatGPT users rather than internal task design alone.
• OpenAI says it has applied LifeVal for a year to measure and improve model performance on these tasks.
• The announcement frames everyday-task competence as part of OpenAI's AGI mission to benefit all of humanity.