Braintrust says web search narrowed the gap between AI models on current-events questions
Braintrust evaluated 1,329 questions across four models and 14 conditions. It reports bigger search gains for more recent events, while a fifth search query was associated with lower performance.
TLDR
Braintrust says it compared You.com, providers’ built-in search and no search on 1,329 current-events questions across four models and 14 conditions. Search reduced the gap between models from 47.9 points to 5.6, it reports. Gains from search declined with event age, from about 45 points for recent events to about 24 for the oldest. Braintrust also reports that runs with five or more searches scored 19–49%, with a fifth query associated with lower performance.
Combined views
17.6K
1 Source, first seen ago