[Image generated with ChatGPT]
[Image generated with ChatGPT]

How many jobs can a company actually cut by adopting AI? As AI-driven organizational restructuring and workforce reductions become a top management priority for companies worldwide, warnings are growing that hasty layoffs can lead to higher costs and a loss of competitiveness. The concern is fueled by a so-called "AI boomerang" phenomenon, in which firms are rehiring for positions they eliminated on the assumption that AI could replace them.

On Thursday, the Associated Press highlighted cases of companies that reduced their workforces citing AI adoption, only to hire back employees afterward. AI has failed to perform as expected in some cases, or has proven unable to replace the experience and institutional knowledge of human workers, prompting companies to revise their staffing plans.

A survey by US staffing firm Robert Half found that 32 percent of hiring managers at American companies that had eliminated roles after adopting AI said they had since rehired for the same or similar positions. The top reasons cited for rehiring were institutional knowledge and work context that AI could not replicate (40 percent), relationship management with customers and other stakeholders (39 percent), and a greater-than-expected need for human oversight and quality verification (38 percent). Some 39 percent also cited increased business demand.

Swedish fintech company Klarna offers a concrete example: after handing customer service over to an AI chatbot to cut costs, the company resumed hiring human agents last year. CEO Sebastian Siemiatkowski acknowledged the company had relied too heavily on the AI chatbot.

US automaker Ford also hired or promoted 350 skilled engineers in June after its AI-based quality-control tools failed to deliver the expected performance. While not a direct case of rehiring laid-off workers, it underscores that skilled human workers remain essential even after AI adoption.

South Korean companies are also reorganizing around AI. SK Networks said in July that, as part of its transition into an AI-focused holding company, it had launched outplacement support, workforce redeployment and a company-wide job redesign initiative. The plan involves identifying where AI can be applied across each division and redefining employees' roles accordingly.

Still, it is difficult to conclude that AI adoption is directly triggering large-scale job losses in South Korea. A comprehensive analysis of AI and the labor market published by the Korea Employment Information Service in August found that 28.6 percent of the 2,297 businesses surveyed had adopted AI. Companies that had done so were found to be using the technology primarily to reduce repetitive tasks, improve productivity and restructure workflows — not to cut headcount. Researchers said AI was simultaneously replacing some tasks, boosting productivity and creating new types of work.

What business leaders need to recognize is that AI's ability to handle individual tasks is not the same as its ability to replace an entire employee. Even if an AI chatbot handles a large share of customer inquiries, skilled agents are still needed to exercise judgment in complex complaints or exceptional situations. And even when document drafting or data analysis is automated, the role of verifying results and taking responsibility for errors does not disappear.

That is why companies need to weigh not just the labor costs saved through workforce reductions, but also the full picture: AI operating costs, quality verification and error-response expenses, the loss of institutional knowledge, and the future cost of rehiring.

Experts also caution against rushing to replace workers. James Landay, co-director of Stanford University's Institute for Human-Centered AI, said at Nvidia GTC this year that judging whether an employee can be replaced based solely on whether individual tasks can be automated risks overlooking the complex, multifaceted nature of real jobs — including collaboration, communication and human relationships. "AI should augment human capabilities, not replace them," Landay said.

Granting AI excessive autonomy also carries risks, experts say. AI specialist Hannah Foxwell said at QCon London 2025, held in the United Kingdom last year, that the engineering required to make AI agents operate reliably in real-world work settings is still in its early stages. Rather than handing entire jobs over to AI, companies should identify which specific tasks can be automated and leave areas requiring creative judgment, problem-solving and customer empathy to people. Foxwell particularly stressed the need for human oversight and intervention to prevent unexpected outcomes.

South Korean companies appear to share this view. In a survey analyzed by the Korea Labor Institute, 84.6 percent of companies using AI agreed that even after adopting the technology, a significant amount of work still needs to be performed by people. Senior research fellow Noh Se-ri said the findings suggest companies are experiencing firsthand how difficult it is to fully replace existing work with AI.

Hasty layoffs can also damage trust between a company and its workforce. Human resources experts interviewed by the AP said employees who are laid off and then offered their jobs back may demand higher pay and greater job security, or press management for explanations of future staffing plans — meaning companies could face additional costs to win back skilled workers.

Ultimately, cutting staff before verifying how much AI can actually replace human workers is a strategy that amplifies business risk. The "AI boomerang" phenomenon emerging overseas suggests that for South Korean companies too, the starting point of an AI transition should not be headcount reduction, but a fundamental redesign of how people and technology work together.


why37@heraldcorp.com