AI's impact has traditionally been assessed in terms of occupations. However, an occupation is comprised of interconnected tasks, and it is these tasks, not occupations themselves, that are affected by AI. To evaluate how tasks may be impacted, previous approaches utilized manual annotations or coarse-grained matching. Leveraging recent advancements in machine learning, we replace coarse-grained matching with more precise deep learning approaches. Introducing the AI Impact (AII) measure, we employ Deep Learning Natural Language Processing to automatically identify AI patents that impact various occupational tasks at scale. Our methodology relies on a comprehensive dataset of 19,498 task descriptions and quantifies AI's impact through analysis of 12,984 AI patents filed with the United States Patent and Trademark Office (USPTO) between 2015 and 2020. Our observations reveal that the impact of AI on occupations defies simplistic categorizations based on task complexity, challenging the conventional belief that the dichotomy between basic and advanced skills alone explains the effects of AI. Instead, the impact is intricately linked to specific skills, whether basic or advanced, associated with particular tasks. For instance, while basic skills like scanning items may be affected, others like cooking may not. Similarly, certain advanced skills, such as image analysis in radiology, may face impact, while skills involving interpersonal relationships may remain unaffected. Furthermore, the influence of AI extends beyond knowledge-centric regions. Regions in the U.S. that heavily rely on industries susceptible to AI changes, often characterized by economic inequality or a lack of economic diversification, will experience notable AI impact.

翻译：传统上，人工智能的影响往往以职业为单位进行评估。然而，职业由相互关联的任务构成，正是这些任务——而非职业本身——受到人工智能的影响。为评估任务可能受到的影响，以往方法依赖人工标注或粗粒度匹配。借助机器学习的最新进展，我们以更精确的深度学习方法取代粗粒度匹配。我们引入AI影响（AII）度量，利用深度学习自然语言处理技术自动识别影响各类职业任务的人工智能专利，并实现规模化分析。本方法基于包含19,498项任务描述的综合数据集，通过分析2015至2020年间向美国专利商标局（USPTO）提交的12,984项人工智能专利，量化人工智能的影响。研究观察表明，人工智能对职业的影响无法以任务复杂度进行简单分类，挑战了"基础技能与高级技能之间的二分法即可解释AI影响"的传统观点。相反，这种影响与特定技能（无论基础还是高级）密切关联——例如，扫描物品等基础技能可能受影响，而烹饪等技能则未必；同样，放射学图像分析等高级技能可能面临冲击，但涉及人际关系的技能则可能不受影响。此外，人工智能的影响力已超越知识密集型区域。美国那些高度依赖易受AI变革影响的产业（通常表现为经济不平等或经济多元化不足）的地区，将承受显著的人工智能影响。