OpenAI Unveils LifeVal, a Benchmark for Everyday Life Tasks
OpenAI has introduced LifeVal, an evaluation that measures how well AI models handle practical tasks from everyday life, such as finding childcare, accessing government services and creating a budget. The company says it used the benchmark over the past year to track and drive model improvement.
- OpenAI introduced LifeVal, an evaluation of AI performance on everyday life tasks.
- Tasks were drawn from the American Time Use Survey, the Multinational Time Use Study and a new survey of 3,995 respondents.
- Example scores range from 90.4% to 100% across five published tasks.
- OpenAI says it improved models on browser use, computer use and connectors tasks over the past year.
- The company links the work to making ChatGPT a more agentic personal assistant.
Why OpenAI built LifeVal
According to OpenAI, more than one billion people use ChatGPT every week. The company says it wants to build AI that can help with the practical challenges of daily life, including finding childcare, accessing government services, creating a budget, preparing for school, modifying recipes, ordering groceries and designing workout plans.
OpenAI argues that people looking for help need more than advice or a chat response. In its words, to be a truly helpful personal assistant, ChatGPT should be able to take actions that actually get things done. LifeVal is meant to measure model performance over time on such tasks, focusing on areas where users wanted the most help from AI and personal agents.
How the tasks were chosen
OpenAI started from two large studies that track how people spend their time: the American Time Use Survey and the Multinational Time Use Study. Together, the company says, they offer a detailed catalog of everyday tasks across different demographics, times of day and seasons.
To find where people would most value AI assistance, OpenAI added a new demographically representative survey of 3,995 respondents. For each task, it asked questions to identify where everyday people most wanted help from a personal assistant. The resulting tasks cover caregiving, health, education, household administration, personal finances, shopping and accessing government services.
Example tasks and scores
OpenAI published several example tasks with scores for its current models:
- Enroll in the right kindergarten after moving: 93.1%
- Manage volunteers ahead of a flood: 98.0%
- Find a new home for my family: 100%
- Watch a game with grandma from far away: 96.3%
- Meal prep for a parent adjusting to a new diabetes diagnosis: 90.4%
In the home search example, the task asked for a home in Luzerne County that fits a budget, has a separate living unit and keeps the commute short. The assistant reported finding one matching home, calculating the transfer tax and adding the deed to a cart for review. In the diabetes example, the output was a first-week plan with meals, reminders and questions for the care team.
OpenAI says its latest models score very highly on many LifeVal tasks.
What it means for ChatGPT
OpenAI says that over the past year it trained its models to become much faster, more capable and more aligned on browser use, computer use and connectors tasks. According to the company, enabling these capabilities in ChatGPT expands the kinds of tasks the model can help with, moving it beyond a chat tool and towards a truly agentic personal assistant.
The company frames LifeVal as part of a broader shift away from academic assessments of model knowledge. It says measuring how well AI helps people navigate decisions, manage everyday responsibilities and get meaningful things done is necessary to ensure AGI benefits all of humanity. OpenAI says it is hopeful that LifeVal represents a step towards that goal.
OpenAI: LifeVal
