机器人和AI模型共同可能承担美国人约80%的工作，Anthropic表示。但机器人仅在0.3%的任务中比人力更便宜。
Robots and AI models together could take on about 80% of the work Americans do, Anthropic says. But robots are cheaper than people for just 0.3% of tasks today.
这一发现源自Anthropic本周发布的新机器人研究，Business Insider对此进行了报道。该研究延续了该公司此前关于AI聊天机器人如何影响就业的研究。
The findings come from Anthropic’s new robotics study, which Business Insider picked up this week. It extends the company’s earlier research on how AI chatbots affect jobs.
不同岗位面临风险：AI模型单独可承担约一半的工作，研究发现。加入机器人后，这一比例升至约81%。
Different jobs at risk AI models alone reach about half of all work, the study found. Adding robots pushes that to about 81%.
两者影响的岗位截然不同。机器人覆盖的是体力劳动且薪酬较低的岗位——而AI聊天机器人则覆盖其他类型，这与通常的AI影响模式相反。
The two hit very different jobs. Robots reach physical, lower-paid roles that chatbots miss, the opposite of the usual AI pattern.
在运输领域，仅有不到15%的任务单独暴露于AI模型之下。加入机器人后，这一比例升至约90%。暴露度最高的十个岗位中有九个是车辆司机，其中出租车司机排名第一。
In transport, fewer than 15% of tasks are exposed to AI models alone. With robots, that jumps to about 90%. Nine of the ten most exposed jobs are vehicle drivers, led by taxi drivers.
成本问题：研究称机器人技术上可处理约74%的体力任务。但这些任务大多仅在受严格控制的环境中可行，如装配线。
The cost problem Robots can technically handle about 74% of physical tasks, the study says. Most of those only work in tightly controlled settings, such as assembly lines.
成本是更大的障碍。目前约有30万美国工人从事的工作中，机器人已比人力更便宜，主要集中在包装和装箱工人。
Cost is the bigger barrier. About 300,000 US workers hold jobs where robots are already cheaper, mostly packers and packagers.
如果机器人价格下降20%，这一数字将升至约280万工人。按过去的变化速度，机器人覆盖10%的任务大约需要40年。园丁呢？
If robot prices fell 20%, that number would rise to about 2.8 million workers. At past rates, reaching 10% of tasks would take about 40 years. What about gardeners?
Business Insider的Alistair Barr质疑这一转变对园艺等动手工作意味着什么。其自身估计显示，清洁和维护工作（包括园丁）在暴露度提升方面将是最大的受益者之一。
Business Insider’s Alistair Barr questioned what the shift means for hands-on jobs such as gardening. Its own estimate put cleaning and maintenance work, which includes gardeners, among the biggest gainers in exposure.
Barr表示持怀疑态度，因为拔除杂草的根部非常困难。Anthropic的研究指出，清洁机器人通常比人工慢几倍，且成本是人工的几倍。
Barr said he was sceptical, given how hard it is to pull a weed out by the roots. Anthropic’s study notes that cleaning robots are often slower than people and several times more expensive.
这些评估是由Claude通过网络搜索得出的，Anthropic强调，暴露度并不等同于岗位流失。该公司还绘制了三种AI经济形态，从温和的就业提升到大规模岗位流失。
The ratings were made by Claude using web search, and Anthropic stresses that exposure does not mean job losses. The company has also mapped three AI economies, from a mild boost to mass job loss.