OpenAI has published LifeVal, an evaluation that grades models on everyday tasks. The tasks come from time-use research and a survey of 3,995 respondents, covering caregiving, health, education, household administration, finances, and shopping. Showcased results include 93.1 percent on shortlisting kindergartens after a move and 90.4 percent on a meal plan for a parent with a new diabetes diagnosis.Should everyday usefulness be the headline metric for AI?#OpenAI #AI #Technology
OpenAI has published LifeVal, an evaluation that grades models on everyday tasks. The tasks come from time-use research ...
