While Data Science for Communication and Social Networks regularly consists of students breaking off into groups to conduct research, in Fall 2024, one group of graduate students’ research proposal was intriguing enough for their teaching assistant to want to eventually publish the project, consulting with the professor of the course in order to do so.

Their labor market study found that when job descriptions align with the capabilities of artificial intelligence, workers are put into a vulnerable position, being offered lower salaries as AI wins a competitive advantage. 

Luca Luceri, the professor for the course and research assistant professor of computer science, said that this group of students was particularly “motivated” and that student groups are not usually expected to publish their research papers.

“This is a great example to showcase how this kind of class project can lead to an outcome and impactful publication,” Luceri said. “It’s a very interesting story to show and motivate students, especially those who want to do research.”

Eun Cheol Choi, the class’s TA  and one of the paper’s lead writers, said the aim of this research was to measure the impact of AI innovation and identify whether recent job descriptions are related to innovation in AI.

“The main thing that I tried to achieve with this framework was to measure the impact of AI technology and specifically the innovations that happen in AI … and see if those innovations can be related to some of the job descriptions and job openings that are popping up every day,” said Choi, a fourth-year Ph.D. candidate at the Annenberg School for Communication and Journalism.

By collecting data on LinkedIn job postings, Choi and his colleagues created an index to measure the relevance and impact of AI innovation in each job sector. They looked into job openings and their related job descriptions, skill requirements and salaries in order to do so.

They found that utilization of AI innovation had the most effect on wholesale, transportation, manufacturing and information sectors. The effects were negative in wholesale, transportation and manufacturing sectors, in which increased effects of AI were associated with lower salaries. On the other hand, AI innovation in the information sector led to higher salaries for employees.

“The size of the change is not homogeneous. Some are impacted more, some are impacted less, and some become more vulnerable to AI,” Choi said.

According to the study, the information sector may have benefited from AI because it serves a role in developing and deploying AI. This contrasts with other sectors’ jobs that are more susceptible to being replaced by AI.

Jobs associated with human-centric tasks — such as legal, human resources and public sector roles — were not as susceptible to influence by AI innovation.

The paper also noted that the job sectors with a higher risk of low wages in the face of AI are also composed of low-income frontline workers, suggesting that exposure to AI poses the threat of exacerbating labor market inequalities and amplifying overall socioeconomic inequality, according to the study.

“It is very important to monitor where those kinds of vulnerabilities in society are located and come up with better interventions that can mitigate those discrepancies,” Choi said.

While Choi and his team had expected such results based on the existing literature on AI posing a threat to blue-collar jobs, he said he hadn’t anticipated this “stark” discrepancy between job sectors, especially with regards to the clear benefits derived from AI in the information sector compared to the clear consequences for the wholesale trade and manufacturing sectors.

John Carlsson, associate professor of industrial and systems engineering at the Viterbi School of Engineering, said that ideally, people will have the chance to “leverage this AI to do things that they couldn’t otherwise.”

He said that nowadays, there are various AI tools that make it easier for people to complete tasks like creating a user interface for an app, empowering people to utilize AI.

Carlsson pointed out some current “barriers to entry” when it comes to AI literacy, including one’s own intimidation and hesitation to get started.

“If it can’t perfectly solve the problem for them, they’re not interested in using it,” Carlsson said. “People are more likely to use it aggressively and seriously when it’s clear to them what problem it can solve for them.”

Carlsson said AI literacy — the ability to understand and utilize AI — is important to “future-proof” one’s job. He said that although the utilization of AI doesn’t meet the needs of various fields, including creative fields like fashion and media, there’s room for it to grow and become a tool.

“AI is not only a harm … but also a useful tool and method to detect some of the changes that are happening in society,” Choi said. “Closely monitoring that with the methodologies that are evolving in this area would be very nice.”