Recycling Batteries
AI OrgReview: The Potential Impact of GenAI on the workforce
Sia Partners has recently developed a solution to assess the potential impact of Generative Artificial Intelligence (GenAI) technologies, such as Large Language Models (LLMs) and Generative Pre-trained Transformers (GPTs), on an organization’s workforce.
Our solution has been developed to assist organizations to strategically plan for the adoption of this new group of technologies and to accurately model the impact on job roles.
We applied this solution, which we have named AI OrgReview, in a proof-of-concept activity to create the outputs detailed in this working paper. In this analysis, we investigated the potential implications of GenAI technologies on a random selection of fifty occupations from across industries. The descriptions for these occupations included roughly 1,000 underlying activities. Applying a scoring rubric (see Figure 1) originally developed by OpenAI and the University of Pennsylvania in a 2023 study examining the potential impact of GenAI on the labor market in the United States, we evaluated these activities based on their alignment with large language model (LLM) capabilities.
Our findings indicated:
- 36% to 46% of reviewed activities (or tasks) were found to be “exposed” to GenAI automation;
- 79% of the occupations assessed could have at least 10% of their work activities impacted by the introduction of LLMs; and
- Approximately 28% of occupations will have 85% or more of their underlying activities impacted, indicating significant change for this portion of the workforce.
Our key finding from the study was that with access to an LLM, at least 36% of all tasks assessed in the sample could be completed significantly faster at the same level of quality. When incorporating additional software and tools built on LLMs - (an example would be querying internal and external data in custom RAG apps), we estimate an upper boundary of 46% of all tasks.
Our analysis suggests that tasks involving information processing and creative work may be particularly susceptible to the impact of GenAI. While this working paper does not systematically examine the characteristics of the most exposed activities and occupations and contrasts them against those at lower exposure levels, we believe that such an exploration would be a valuable direction for future research.
