Artificial intelligence (AI) is erasing the boundaries of work hours for high-income professionals. Data reveals a growing divide in the labor market, where skilled workers who delegate tasks to AI enjoy greater optimism about their wages and job security, while entry-level workers face a hiring cliff.
According to the Cadences report by Anthropic, an analysis of usage patterns for the AI chatbot Claude from April 10 to June 10 showed that high-paying jobs in fields like marketing and software development receive a steady stream of work requests even during nights and weekends.
High-Income Workers Shift to 'Work Mode' During Off-Hours
The report indicates that inquiries related to news peak at 7 a.m., while requests for drafting work emails surge between 10 and 11 a.m. Requests for cooking recipes spike to 2.3 times the usual volume at 6 p.m., and sleep consultations are concentrated around 5 a.m. In the U.S., tax-related conversations can increase by up to eight times just before tax filing deadlines. This illustrates a typical pattern where the type of tasks delegated to AI changes throughout the day.
High-income workers, however, are deviating from this pattern. An analysis based on the U.S. Bureau of Labor Statistics (BLS) wage quartiles shows that the share of work-related conversations during nights and weekends increased by 8% for the top half of earners (third and fourth quartiles), while it decreased by 4% for the lowest quartile and 11% for the second quartile. Even when excluding the computer and math fields, which often require late-night work, the trend of increased AI usage among high earners remains.
On weekends, personal use of Claude accounts for 50% of conversations, up from 35% on weekdays, indicating that most users take a break from work. However, high-income professionals continue to engage in work-related discussions during these hours. The token usage for work conversations among the highest earners is approximately 2.07 times that of the lowest earners, suggesting that AI is beginning to blur the lines between work and personal life, with high-income workers at the forefront.
Optimism Among AI Users, Anxiety for Junior Workers
Notably, there is a correlation between how AI is utilized and future job outlooks. An analysis of nearly 9,700 respondents linked to actual Claude usage records revealed that users who delegate tasks to AI report higher job satisfaction and optimism about wage increases and job stability over the next year.
Workers who possess the skills and discretion to delegate tasks to AI view it as a leverage rather than a threat. In fact, 93% of Claude conversations resulted in tangible outputs such as explanations, reports, or code.
Respondents reported specific benefits: 86% said their work speed increased due to AI, and 82% noted an expanded range of tasks they could handle. Additionally, 57% believed their skill value had risen, a sentiment that was even stronger among those who frequently delegate tasks to AI.
Contrary to the belief that delegating tasks might reduce learning opportunities, 68% of users who delegate frequently reported learning more from AI, similar to other users.
In contrast, early-career workers expressed significant anxiety about job security, fearing that AI could replace their roles. While only 10% of all respondents believed they would lose their jobs within a year, over one-third of junior colleagues estimated a 60% or higher chance of job loss.
Decline in Employment for Young Workers: An 'Invisible Hiring Cliff'
The anxiety among entry-level workers is not merely psychological; it is supported by actual salary data. A research team led by Professor Erik Brynjolfsson at Stanford University's Digital Economy Lab extended an analysis of payroll data from ADP, which covers one in six U.S. workers, through April of this year. They found that employment in high AI exposure jobs decreased by only 0.2% across all age groups. However, for those aged 22 to 25, the decline was 3.8%, with the rate of decrease accelerating by about 0.5 percentage points each month.
In contrast, employment for those aged 35 to 40 in high AI exposure jobs increased by 6% to 9%. Companies are opting to halt new hiring rather than lay off existing employees as they transition to AI, creating an 'invisible hiring cliff' that is not reflected in overall employment statistics.
Statistics from the U.S. reemployment consulting firm Challenger, Gray & Christmas indicate a similar trend. In their July layoff report released on August 6, AI was cited as the leading reason for layoffs for five consecutive months. Of the 33,429 layoffs announced in July, 33% (10,970) were attributed to AI. From January to July of this year, AI was mentioned in 112,713 layoffs, accounting for about 24% of the total.
Particularly, layoffs in the tech sector have surged by 67% year-over-year, totaling 149,023, representing 31% of all layoffs. However, new hiring plans during the same period increased by 25% compared to the previous year. While Challenger's analysis suggests that AI is not dismantling the labor market but rather reshaping it, the costs of this transition are disproportionately affecting specific generations.
These data indicate that the fractures created by AI in the labor market are emerging not only along traditional lines of 'high income versus low income' but also along the career stage divide between those who can delegate tasks and those who cannot.
Skilled workers who adeptly utilize AI gain efficiency, job satisfaction, and positive income outlooks, but they are sacrificing their evenings and weekends. Ironically, their increased productivity is erasing job opportunities for newcomers. Concerns are growing that the elimination of entry-level positions could disrupt the supply of mid-level management talent in the long term.
South Korea's High AI Adoption Yet Lack of Labor Market Data
The challenge is that South Korea lacks the data to measure this divide. The Cadences report ranks South Korea 14th among 121 countries in terms of Claude usage, with a usage index 3.78 times higher than expected based on population and economic size, indicating the highest level of AI adoption globally.
Given the rapid pace of AI adoption, the labor market changes observed in the U.S. could quickly manifest in South Korea. However, the national employment statistics do not differentiate AI exposure by job type, and discussions surrounding the AI Basic Act and government policies focus on industrial development and infrastructure investment, leaving a significant gap in tracking the impact of AI on the domestic labor market.
While the U.S. can observe the hiring cliff for young workers in real-time through private payroll data, South Korea lacks the means to confirm similar trends.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.