- Generative AI could lead to the widespread automation of jobs, new research found.
- But blue-collar roles in manufacturing and construction face minimal influence, data indicates.
Rapid advances in artificial intelligence are stoking employee concerns about the automation of jobs.
Generative AI, a type of AI capable of generating text and other content in response to user prompts, has exploded in popularity in recent months following the launch of OpenAI's ChatGPT. Some companies have already tapped the chatbot to take on content-producing roles.
A recent report from Goldman Sachs found the tech could significantly disrupt the labor market, with it affecting about 300 million full-time jobs or 18% of work. Another recent study, from OpenAI with the University of Pennsylvania, found OpenAI's AI chatbot ChatGPT could affect around 80% of jobs in the US. White-collar roles could be among the worst hit, the study also found.
While the studies predicted advances in AI tech could also improve labor productivity and create jobs, they said some industries would likely be more affected than others.
Legal services
The legal industry has been cited in multiple studies as one of the industries most exposed to automation.
A study released in March by researchers at Princeton University, the University of Pennsylvania, and New York University, estimated that legal was the industry most likely influenced by AI tech like ChatGPT. The researchers used a benchmark that matched specific work tasks with AI capabilities to calculate the results.
Manav Raj, one of the authors of the study and an assistant professor of management at the Wharton School of the University of Pennsylvania, told Insider this was because the industry was made up of a relatively small number of occupations already exposed to automation, such as legal assistants.
The report from Goldman Sachs, which relies on an analysis of data on occupational tasks in the US and Europe, also highlighted legal workers as especially at risk from the new tech.
University education and research
Education was one of the first areas to witness some of the effects of ChatGPT, and institutions, especially those in higher education, are scrambling to deal with the consequences of the technology.
The study led by Princeton University, which was authored by Raj, Ed Felten, and Robert Seamans, identified postsecondary teachers as some of the workers highly exposed to large language models.
Professors across a variety of subjects made up eight of the top 10 affected occupations, while junior colleges were ranked the 17th most likely sector to be influenced. Raj said this was because AI chatbots tended to be very good at research tasks such as identifying patterns and writing reports.
The research from Goldman Sachs, which was written by Joseph Briggs and Devesh Kodnani, estimated almost 30% of work tasks in education instruction and library services in the US and Europe could be automated.
Administration
Administrative roles have particularly high exposure to automation, Goldman researchers found. They estimated that about 46% of work tasks in the sector could be automated.
Wharton's Raj said that sectors like administration, where there's a "relatively high prevalence of white-collar roles, tended to be more exposed" to advanced AI technology. But the effect of AI on the labor market will be "multifaceted," Raj said, in some cases complementing human work.
Finance
Banks are already incorporating AI tech into their day-to-day business operations.
More than half of banks say they've used the tech for revenue generation, according to the Cambridge Centre for Alternative Finance and the World Economic Forum. Financial quantitative analysts were also identified in a study led by OpenAI and the University of Pennsylvania as one of the occupations highly exposed to advanced AI technology.
Tech
AI chatbots like ChatGPT have excelled at writing code. The new programs can analyze numbers and produce code, often at a faster rate than humans can.
The study led by Princeton University ranked software publishers as the 11th most likely sector to be affected by tech like ChatGPT, something Raj put down to the chatbot's impressive coding abilities. The Goldman researchers estimated that about 29% of computing and mathematical tasks in the US and Europe could be automated.
Construction
Sectors with a high number of blue-collar roles, such as construction, scored relatively low on Goldman Sachs' report. In the US and Europe, the researchers estimated less than 10% of work tasks in the sector were exposed to automation.
Construction and related workers were also ranked some of the least influenced occupations on the list compiled by researchers at Princeton University, the University of Pennsylvania, and New York University.
Trade workers
Trade workers are also projected to be relatively untouched by the tech.
Trade work requires practical, hands-on experience and specialized training, and many of the tasks involve manual labor.
Craft and related trade workers had the lowest share of industry employment exposed to automation in Europe, Goldman Sachs found. The researchers estimated that only 4% of the industry was exposed to automation by AI in the region.