AI’s True Cost: Betsy Tong Highlights Hidden Downsides of Automation

In a recent LinkedIn post, Betsy Tong discusses the often-overlooked negative consequences of implementing Artificial Intelligence (AI) in the workplace, arguing that initial cost savings can mask significant long-term detriments to quality and operational efficiency. Tong challenges the prevalent narrative that AI solely equates to productivity gains through workforce reduction, presenting case studies that illustrate the pitfalls of this approach.

The Illusion of AI-Driven Productivity

Betsy Tong begins by framing AI’s promise against its perceived reality in many executive suites. Instead of lightening the workload, Tong observes that AI has frequently been leveraged to reduce payroll, a move often rebranded as a productivity enhancement. This perspective is starkly illustrated by examples from major companies.

“Execs used it to lighten payroll instead. Then called it productivity.”

Tong highlights the case of Klarna, which reduced its workforce significantly, attributing the capability of 700 customer-service agents to AI. However, this strategy proved short-lived. As Tong points out, Klarna began rehiring humans in May 2025, with CEO Sebastian Siemiatkowski admitting to Bloomberg, “As cost unfortunately seems to have been a too predominant evaluation factor when organizing this, what you end up having is lower quality.” This reversal underscores Tong’s central argument: prioritizing cost savings over quality can lead to a detrimental cycle.

Unforeseen Operational Costs

The article delves into the hidden costs that are often absent from initial AI impact reports. Tong argues that while the reduction in headcount is an easily quantifiable saving, other critical factors are frequently ignored.

• Lost institutional expertise
• Lower quality output
• Increased need for human intervention on exceptions
• Employees working longer hours to compensate for AI shortcomings

Commonwealth Bank serves as another example in Tong’s analysis. The bank claimed its AI bot reduced call volumes, leading to the elimination of 45 customer service jobs. However, according to Tong, the union contested these figures, reporting that call volumes were increasing and staff were overworked. The bank eventually reversed its decision, stating the roles “were not redundant.” This pattern, Tong suggests, reveals a fundamental “design gap” in how AI is implemented.

“If your AI productivity gains only appears when you leave quality out, it is cost shifting.”

A study tracked by UC Berkeley Haas at a U.S. tech company for eight months further supports Tong’s thesis. The research indicated that while AI might make employees work faster, it also intensified their workload, requiring higher output and increased handling of exceptions rather than improving overall efficiency.

The Real Leadership Test

Betsy Tong posits that the true measure of leadership in the age of AI is not the ability to replace people with machines, but rather to ensure the business performs better with humans in the loop, rather than in their absence. She proposes a practical test for businesses to gauge their actual AI productivity.

Measuring True AI Productivity

Tong suggests pulling the previous quarter’s corrections and counting how many required human intervention. This, she argues, provides a more accurate measure of AI’s impact than simple headcount reduction.

“The real leadership test is not whether AI can remove people. It is whether the business performs better if humans are gone.”

The data reinforces this point, with a Robert Half report indicating that 32% of companies that cut staff for AI have since rehired for those same roles, a figure rising to 44% in the finance sector. Tong concludes by urging leaders to reframe their AI productivity Key Performance Indicators (KPIs) to reflect a more holistic view of performance that includes quality and operational sustainability, not just immediate cost reduction.