OpenAI's new LifeVal benchmark tests AI on everyday life tasks

OpenAI has built a new evaluation called LifeVal that measures how well its models handle the practical tasks people deal with day to day, rather than academic knowledge tests. The goal is to help the more than one billion weekly ChatGPT users get real assistance with things like finding childcare, accessing government services, budgeting, preparing for school, modifying recipes, ordering groceries and designing workout plans.

To build the benchmark, OpenAI started from existing time use research, the American Time Use Survey and the Multinational Time Use Study, which catalog everyday tasks across different demographics, times of day and seasons. It added its own survey of nearly 4,000 people asking where they most wanted help from a personal assistant, covering caregiving, health, education, household administration, personal finances, shopping and government services.

OpenAI says it has used LifeVal over the past year to track and improve how its models perform on these tasks, and that its latest models score very highly on many of them, citing examples like finding the right kindergarten after a move or planning meals for a parent with a new diabetes diagnosis. The company frames this work as part of a broader push to make ChatGPT act more like an agentic personal assistant that can take actions, not just offer advice.

Why it matters

It gives a concrete way to judge whether AI assistants are actually useful for ordinary life admin, not just trivia and coding.

Source: OpenAI