AI 不再只是聪明，还能像最靠谱的员工一样持续执行复杂目标——做 AI 应用或自动化工具的开发者，值得思考如何利用这种超人类尽责性来设计下一代产品。
Timothy Bates 指出，AI 在去年 11 月左右已达到专家级通用智能，现在又具备了超人类的尽责性。尽责性使 AI 能够像人类一样设定高性能目标：策略制定、管理复杂性、招募资源、保持注意力、坚持并获取反馈，直到目标完成。工具如 Cursor 2.5 和 Antigravity 2.0 将模型智能嵌入计算时间框架，重现了这套系统。AI 没有挫败感，可以持续工作数年甚至数百年，解决极其困难的任务。这标志着 AI 从单纯智能迈向自主执行复杂长期任务的新阶段。
Interesting.
Interesting. Timothy Bates @timothycbates Around November, AI reached expert-level broad intelligence. Now, this is coupled with super-human conscientiousness. For us, conscientiousness embeds our intelligence and capacity for labour in a system for high performance goal-setting: Strategy, Management of complexity, Recruitment of resources, Attention, Persistence, Acquiring feedback – all operating until goal accomplishment. Tools like @Cursor 2.5 and @antigravity 2.0 embed model intelligence in compute-time harnesses recreating this same system for high performance goal-setting: strategizing, getting feedback based on checking simulated and real outcomes, attending to tasks, recruiting resources, dividing tasks to manage their complexity. The complete absence of frustration means it can toil for years, perhaps centuries (!) on unimaginably difficult tasks. What a time to be creative! 🔗 View Quoted Tweet 💬 6 🔄 2 ❤️ 15 👀 6978 📊 6 ⚡