AI is Already Disrupting Employment - The "Seismic Shift in Jobs" Reflected in Statistics

Will AI take jobs away, or will it assist human work while creating new roles? Until now, this debate has often been discussed as a "future talk" driven by expectations and fears. However, an article published by Phys.org on April 16, 2026, indicates that we are moving beyond that stage. The article is based on an analysis by researcher Clinton Free, contributed to The Conversation, which discusses the possibility that the restructuring of the labor market by AI may have already begun, using U.S. employment data as a clue.

What makes this analysis significant is that it does not resort to extreme alarmism like "an era of mass unemployment is coming," but instead examines where changes might first appear, categorized by job type. Notable is the vulnerability of jobs involving a lot of routine information processing, such as customer support, administration, software, and IT services. Additionally, there is a gradual slowdown observed in marketing, banking, travel, and retail. The author perceives these changes not as a "collapse," but as a "targeted restructuring." In other words, AI is not eliminating all jobs at once but is beginning to change hiring and placement, starting with white-collar jobs that involve a lot of routine work.

More unsettling might be the "stagnation" rather than flashy layoffs. In professions that have supported modern large corporations, such as finance, consulting, management, and corporate support, long-term expansion is said to be slowing. If companies determine that AI can take over tasks like document creation, summarization, initial analysis, report writing, and inquiry handling, they can significantly reduce labor costs by simply limiting new hires, even without immediately dismissing existing employees. Changes in employment appear as "disappearance of job postings" before layoff notices. In this sense, the impact of AI is more likely to manifest in the job market before affecting unemployment rates.

The most affected by this shift are likely to be the inexperienced younger generation. The Phys.org article points out that the unemployment rate for recent college graduates in the U.S. is about 5.6%, about 7% for young graduates, and 42.5% are in "underemployment," working in jobs that do not require a degree. The New York Fed's special page also continuously tracks the unemployment and underemployment of recent graduates, treating the recently graduated group as early-career individuals aged 22-27. In short, the impact of AI is more likely to manifest as "reducing opportunities for young people to get their first job" rather than "taking jobs away from veterans." This is more serious than a mere economic downturn because it narrows the very entry point for gaining experience.

However, not all jobs are weakening in the same way. The article states that blue-collar jobs such as construction and maintenance are relatively strong, with blue-collar employment growth surpassing white-collar growth over the past three years. While tasks like writing, summarization, analysis, and inquiry handling are becoming AI's forte, jobs that require complex on-site judgment, physicality, and quick adaptation to environmental changes are still difficult to replace. The long-held belief that "obtaining an education and working in an office" is the path to stability is quietly beginning to crumble.

What makes this change troublesome is that its speed and scope are greater than previous technological innovations. The original article cites that ChatGPT reached over 100 million users within two months of its release and that AI is penetrating not only simple tasks but also the core of cognitive labor, such as legal documents, code, financial materials, and marketing texts. There is a history of general-purpose technologies like steam engines and computers changing society. However, this time, the adoption is fast and permeates across finance, law, logistics, hospitality, and management departments. Therefore, both companies and workers have little room to adapt slowly.

However, what is important here is not to stop thinking with "AI is scary." In fact, a Gallup survey in February 2026 found that half of U.S. workers use AI at least a few times a year for work, with 13% using it daily and 28% using it more than a few times a week. 41% said their workplace has introduced AI tools for organizational operations. The same survey introduced by AP News shows that many people who use AI feel an increase in productivity, while those who do not use it keep their distance due to distrust in accuracy, ethical concerns, and data privacy. In other words, the reality in the workplace is neither "everyone is saved by AI" nor "everyone loses their job," but rather benefits and caution are progressing simultaneously.

The on-the-ground experience is also not straightforward. The Guardian reported that while management emphasizes productivity improvements through AI, workers feel a "workslop" sensation, being overwhelmed with correcting AI-generated errors and fixing low-quality outputs. AI certainly speeds up some tasks, but that does not necessarily translate into overall workplace efficiency improvements or a good work experience. If the reduction in personnel increases the checking burden on those who remain, statistical efficiency may simply become pressure for the field.

This sentiment is clearly reflected on social media as well. On LinkedIn, the original article's author himself posted about the contraction and stagnation in customer service, administration, finance, marketing, and parts of software, particularly softening in entry-level roles, and the issue raised by the article was widely shared among professionals. Meanwhile, on platforms like Reddit, the discussion is more raw. The central anxiety is not about "jobs becoming zero," but rather the sense that "the value of labor is decreasing, and young people's bargaining power is being lost." There is strong caution about AI reducing wages and the initial momentum of careers.

However, there is also strong caution on social media. While it is true that AI is distorting the hiring market, it is premature to explain everything solely by AI. In fact, on Reddit, there is a discussion that "AI is not immediately destroying employment itself, but rather companies are changing their hiring methods," and in another thread, there are voices saying, "Recent graduate employment is tough, but the story is more complex." Given that economic slowdown, corporate hiring restraint, post-COVID backlash, and stricter job requirements are overlapping, AI is one of the main factors but not the sole cause. This reservation is necessary to keep the discussion calm.

Nevertheless, it is undeniable that the sentiment of the younger generation is cooling. A Gallup survey introduced by The Verge shows that among the 14-29 age group, "hope" and "excitement" for AI have decreased compared to the previous year, while "anger" has increased. Nearly half believe that the risks of using AI in the workplace outweigh the benefits, and 80% see that learning in the future will be impaired as AI becomes more efficient. They are not rejecting AI. Rather, they use it because it is necessary. However, distrust and fatigue are beginning to mix with that necessity. Here lies the core of the current employment anxiety. They use it because it's convenient. But that convenience might be eroding their future.

It is too early to definitively declare the end of employment due to AI at this point. However, the signs that the entry point of employment, the initial career of new graduates, and routine-heavy white-collar jobs are being shaken first are becoming harder to ignore. The question to ask is not "Can we stop AI?" but "Who will bear the cost of change?" Income support, retraining, curriculum reform, maintaining young recruitment, and transparency in AI implementation. Accelerating technology without such institutional design will first cut off those who have neither achievements nor bargaining power. The current employment statistics show precisely how that entry point is narrowing.

Source URL

・Phys.org article (Reposting Clinton Free's analysis from The Conversation, discussing early signs of employment restructuring by AI using U.S. employment data)
https://phys.org/news/2026-04-employment-early-ai-job-disruption.html

・New York Fed "The Labor Market for Recent College Graduates" (Basic information on the definition and tracking data of the unemployment and underemployment rates of recent graduates)
https://www.newyorkfed.org/research/college-labor-market

・Gallup "Rising AI Adoption Spurs Workforce Changes" (Survey on AI usage rates among U.S. workers, AI implementation rates at workplaces, and productivity perceptions)
https://www.gallup.com/workplace/704225/rising-adoption-spurs-workforce-changes.aspx

・AP News "Why some workers are embracing AI while others won’t use it, according to a new Gallup poll" (Summary of AI utilization and caution in the workplace based on a Gallup survey)
https://apnews.com/article/e4c129e9773255203ccae208bfccb367

・The Guardian "I feel helpless: college graduates can't find entry-level roles in shrinking market amid rise of AI" (Report on new graduate employment difficulties and underemployment)
https://www.theguardian.com/us-news/2026/apr/12/college-graduates-job-market-ai

・The Guardian "Bosses say AI boosts productivity - workers say they're drowning in 'workslop'" (Report on on-site burden and "workslop" sensation after AI implementation)
https://www.theguardian.com/technology/2026/apr/14/ai-productivity-workplace-errors

・The Verge "Gen Z’s love-hate relationship with AI" (Introduction of changes in young people's feelings about AI, decrease in hope, and increase in anger)
https://www.theverge.com/ai-artificial-intelligence/909687/gen-z-doesnt-like-ai-gallup

・Clinton Free's post on LinkedIn (Public post sharing the key points of the original article on social media)
https://www.linkedin.com/posts/clinton-free_employment-data-shows-the-early-signs-of-activity-7450308769512845312-9Bsk

・Reddit discussion 1 (Example of anxiety about AI not eliminating jobs but rather damaging labor value and bargaining power)
https://www.reddit.com/r/Futurology/comments/1sc35bm/economists_once_dismissed_the_ai_job_threat_but/

・Reddit discussion 2 (Example of cautious opinion that while AI has an impact, changes in the hiring market and other factors are also significant)
https://www.reddit.com/r/jobs/comments/1snkxer/ziprecruiter_says_grad_job_market_is_improving/
https://www.reddit.com/r/ArtificialInteligence/comments/1qu6csk/is_ai_actually_destroying_jobs_or_are_we/