Background

In a recent effort to gauge the future influence of large language models (LLMs) on employment, AI researchers gathered expert projections on how LLM‑powered software might alter job tasks. The study deliberately refrains from predicting when such technologies will be widely adopted, opting instead for an open‑ended horizon that prioritizes theoretical potential over concrete timelines.

Methodology

Experts were asked to evaluate the "theoretical capabilities" of LLMs across a broad spectrum of occupations. The analysis incorporated forward‑looking assumptions about software that could be built around LLMs, labeling tasks based on the likelihood of being accelerated or transformed by these tools. The researchers explicitly noted that they do not set a self‑imposed deadline for when these effects would manifest, acknowledging the uncertainty surrounding development and adoption rates.

Key Findings

The most striking projection is that between 47% and 56% of all tasks could eventually be made at least 50% faster by leveraging LLM‑powered solutions. Additionally, the study found that 19% of workers occupy roles where over half of their tasks are labeled as exposed to LLM impact. Certain occupations—namely mathematicians, writers and authors, and web and digital interface designers—were identified as "fully exposed," meaning that 100% of their job‑related tasks could be significantly affected.

Examples of potential applications include using LLMs to mediate negotiations by transcribing each party’s perspective and feeding the data to an LLM to help resolve disputes. However, researchers acknowledged that widespread adoption of such tools would require broad acceptance, noting that "many people would need to buy into using new technological tools to accomplish this." They also stressed that accelerating a task with LLM assistance is not synonymous with fully automating or replacing human labor.

Implications

The findings highlight a substantial, though theoretical, capacity for LLMs to enhance productivity across many sectors. While the projected speed gains suggest notable efficiency improvements, the study cautions against interpreting these numbers as forecasts of job elimination. Instead, the research frames LLMs as augmentative technologies that could reshape how tasks are performed, potentially redefining skill requirements and workflow designs in affected occupations.

Used: News Factory APP - news discovery and automation - ChatGPT for Business