哈佛与Perplexity最新研究：AI代理平均每次会话自主工作26分钟，而搜索仅33秒
英文摘要
A joint study by Harvard and Perplexity analyzed 10,000 matched session pairs from Perplexity Search and the AI agent Perplexity Computer over a 90-day window. Computer performed 26 minutes of autonomous work per session (median 9 minutes), a 48× increase over Search's 33 seconds (median 14 seconds). On matched tasks, Computer plus human reduced estimated time by 87% and cost by 94% versus Search plus human, with a meaningful dissatisfaction rate of 1.3% compared to 2.9% for Search. Computer queries also expanded task scope: cross-occupation share rose to 59% (vs 50%), higher-order Bloom's cognition was required in 76% of queries (vs 55%), and 23% of queries addressed task statements never submitted to Search.
中文摘要
哈佛大学与Perplexity联合研究分析了来自Perplexity Search和AI代理Perplexity Computer在90天内的一万对匹配会话。Computer每次会话自主工作26分钟（中位数9分钟），是Search的33秒（中位数14秒）的48倍。在匹配任务上，Computer加人类相比Search加人类将预估时间降低87%，成本降低94%，且其有意义的不满意率仅为1.3%（Search为2.9%）。Computer的查询还拓展了工作范围：跨职业查询占比升至59%（Search为50%），76%的查询需高阶认知（Search为55%），并且23%的查询解决了从未提交给Search的任务描述。
关键要点
AI agent (Computer) delivers 48× longer autonomous work per session than conversational search (26 min vs 33 sec).
AI代理（Computer）的每次会话自主工作时间是对话式搜索的48倍（26分钟 vs 33秒）。
Computer + human reduced task time by 87% and cost by 94% compared to Search + human on matched tasks.
在匹配任务上，Computer加人类相比Search加人类将任务时间降低87%，成本降低94%。
Computer achieved a 55% lower meaningful dissatisfaction rate (1.3% vs 2.9%), despite higher autonomy.
尽管自主性更高，Computer的有意义不满意率反而降低55%（1.3% vs 2.9%）。
Computer queries expanded both horizontally (59% cross-occupation vs 50%) and vertically (76% higher-order cognition vs 55%).
Computer查询在横向（跨职业占比59% vs 50%）和纵向（高阶认知需求76% vs 55%）上均实现扩展。
23% of Computer queries addressed task statements that the same users never sent to Search, indicating newly attempted work.
23%的Computer查询解决了相同用户从未向Search提交过的任务描述，表明尝试了全新工作。