Unverified60% confidenceBenchmarkExact time
Stanford University's WebArena benchmark, released in 2023, showed that even the most advanced AI models achieve only about a 14% success rate on complex operational tasks on real web pages
2
Sources
60%
Confidence
Long-term
Relevance
7/18/2026
First Seen
Sources
Giving AI Root Access to a Computer and a Bank Card: A Full Breakdown of the Autonomous Money-Making Experiment
Related Entities
Related Claims
UnverifiedAI智能体操作网页的核心技术难题是actuation（执行操作），源于AI模型的感知-行动鸿沟69% similarUnverified斯坦福大学2024年AI指数报告显示，编程类任务是AI渗透速度最快的工作领域之一68% similarUnverifiedAccording to industry annual report data, AI alone is essentially unable to complete tasks, but pairing it with a real person improves task completion rates by more than 70%.66% similarUnverifiedAI代理在面对动态加载页面、验证码、多因素认证等情况时仍存在较高的失败率66% similarUnverifiedAI的核心优势在于处理复杂高难度任务而非当作搜索引擎使用66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/680680API
curl https://kongchang.com/api/v1/knowledge/claims/680680MCP
get_claim(id=680680)