March 25, 2026
What does the post-AI job market hold for current and future graduates? News outlets have posed this question many times since ChatGPT’s launch in 2022, and despite the constant and often conflicting headlines, no one has a definitive answer. SCI’s own Assistant Professor Morgan Frank, who works in the Department of Informatics and Networks Systems (DINS), set out to answer this question with findings from his time as a senior fellow at Microsoft’s AI Economy Institute.
The talk, presented by DINS, hosted over 40 attendees across SCI faculty, staff, and master's and PhD students eager to understand national data surrounding the post-AI job hunt.
Frank began with a recap of recent headlines from major news outlets: “You can go back and find among these noteworthy venues contradictory headlines occurring at almost the same time.” Varying reports from trustworthy media sources can add to the cultural atmosphere of uncertainty surrounding the level of competition new grads will face from AI, said Frank.
According to his research, different job types have received different levels of AI-exposure, which is the share of workplace activities that can be automated by large language models. The suddenness of AI’s introduction, its immediate market disruption, made it “one of the, if not the most fast-growing technological services ever introduced.”
Frank, however, sees this early stage of AI technology as an opportunity: his research aims to make predictions about its future with the data collected and answer crucial questions about observed changes in hiring, job responsibilities, and AI-inclusive job preparation. Some of the answers, he found, came from an aggregation of each U.S. state’s unemployment insurance program data.
Looking at the last 10 years of unemployment data, “unemployment risk rose, especially for AI-exposed workers around the launch of ChatGPT,” noted Frank, “but it started before ChatGPT was widely available to the public….broadly across the economy, across a lot of white collar occupations…people are having a harder time securing stable employment, but the timing's off to blame it on ChatGPT.”
Frank then analyzed data from millions of LinkedIn profiles to determine how long it took graduates to land a post-grad job, based on their graduation year and major, to account for potential misrepresentation of recent college grads who do not qualify for unemployment benefits.
For the graduating classes entering a pre-AI job market, the length of the job hunt followed a “really regular pattern that persisted for almost a decade, and our data is no longer applicable for the classes of 2021, 2022, and 2023,” said Frank. “In the months after May of their graduation year, they underperform at securing Al-exposed jobs compared to earlier graduation cohorts.”
But Frank once again questions the narrative promoted by the headlines. He indicates on the graph the points at which 2021 and 2022 grads’ job hunts were impacted by the launch of AI. “You can see that the underperformance for these graduates occurs again to the left of these lines…starting months before the launch of ChatGPT.”
The AI Economy Institute’s findings do support the conclusions that workers in AI-exposed roles are experiencing higher unemployment in the post-AI era, and that recent college graduates are struggling more to secure these jobs compared to those who entered the workforce before AI emerged. Their findings also demonstrate, though, that these shortcomings could be observed on a national scale months before ChatGPT was widely available to the public, so the cause may not be specifically attributed to ChatGPT’s launch.
The answers to these questions still elude Frank’s team. So, what are universities to do? The AI Economy Institute partnered with the Open Syllabus project to find some answers.
Frank and his team analyzed thousands of college syllabi from before and after COVID, as well as before and after the introduction of large language models, to understand what skills academic courses included before the influence of remote learning and tools like ChatGPT. They used natural language processing to scan each syllabus and extract terms aligned with classifications from the US Bureau of Labor Statistics. These findings were then combined with LinkedIn data to examine how the presence of AI-related skills in coursework correlated with individuals’ salaries and job search duration.
Should colleges continue teaching AI-related skills? What should students learn that can boost their employability in a post-AI world, if not AI skills?
“People who started their first job before the launch of ChatGPT: there's basically no relationship between exposure to AI skills and salaries, or to time spent job seeking. However, after the launch of ChatGPT, there's a positive association with salary, and students who are exposed to AI-related skills actually spend less time job seeking compared to their peers,” said Frank. “And doubling down on these skills might be the best service we can provide to our students to maximize their salary and help them in a tough job market.”
Frank’s findings point to the conclusion that the post-AI job market rewards those who can work with AI, not against it. Integrating AI fluency into university education as a core competency can help better equip graduates to stand out and succeed in a shifting job market.
“The future of work tact for this kind of question,” Frank advises, “is to say there have been tons of technological advancements. Think about the industrial revolution, think about natural language processing. And yet we've always had jobs, right? Even though there were these big shifts. That's the classic wisdom in this space is that technology creates new opportunities, creates new industries, and creates lots of new jobs as a result."
View a recording of Frank’s talk, "The Al Effect on Early Careers: Sorting Fact from Hype".