Jobs of Tomorrow: Large Language Models and Jobs | World Economic Forum

This new 34-page report discusses the direct impact of Generative AI, specifically Large Language Models (LLMs), on various jobs and tasks. Based on an analysis of 19,000 tasks in 867 occupations, researchers assessed the level of risk exposure of tasks and jobs to automation via LLM using four categories: 1) High potential for automation: Going forward, the task will be performed by LLMs, not humans. 2) High potential for augmentation: Humans will continue to perform the task, and LLMs will increase human productivity. 3) Low potential for either automation or augmentation: Humans will continue to perform the task with no significant impact from LLMs. 4) Unaffected. Non-language tasks, such as those that emphasize physical movement (e.g., loading products for transport). Table 1 on page 8 summarizes specific tasks and the four categories in which they fall. For example, tasks with the highest potential for automation by LLMs tend to be routine and repetitive, such as administrative or clerical activities, whereas tasks with lower potential for automation and augmentation are those that require a high degree of personal interaction and collaboration (e.g., negotiation of contracts, scientific and technical work). Organizations can use this four-category framework to help break down roles into tasks, and then use this information to determine the most optimal way to accomplish those tasks through different tactics (e.g., automation, full-time employees, external partnerships, etc.). To supplement this report, I am resharing the World Economic Forum’s 2023 Future of Job Report, which explores how jobs and skills will evolve over the next five years.