Story

On air. The free wire runs 15 minutes behind live. See pricing

World, crypto, finance and tech news, scored for impact.

AITokenPostKorean, translatedFiled 18 h ago

Search sessions up 0.23%... screen delays also worsen in Meta AI experiment

Original headline검색 세션 0.23% 증가…메타 AI 실험서 화면 지연도 악화

Open the sourceReported by TokenPost, tokenpost.kr, 20 November 2026, by 김민준

Not investment advice. Muhabbit scores how much a story may matter, not which way to trade.

Meta researchers reported that an AI agent-designed search suggestion model, tested in a 31-day campaign, lifted search sessions by a relative 0.23% versus a control group, though the time to render the search screen increased. The model, called Nullstate, orders the suggestions users see before they type in the search box, and the paper cites processing volumes exceeding 1 billion requests per day—a model-request metric distinct from user or session counts. The AI agent revised model code, ran training jobs, and performed offline evaluations, with the researchers using Claude Opus 4.8. The result matters as an early signal that agent-driven model iteration can move a core engagement metric at Meta's search scale, while the slower screen rendering flags a latency trade-off to watch. Next, investors should watch whether Meta expands the experiment and whether the session gain holds once rendering delays are addressed.

Muhabbit's AI wrote this summary in its own words from TokenPost's report and translated it from Korean. It can get details wrong, so check the source.

Named in this story

Meta (META), Nullstate, Claude Opus 4.8

A linked name opens that asset's sentiment page.