In 2023, IBM dismissed nearly 8,000 employees, primarily within its human resources division. It was a high-profile signal of the company’s ambition to automate routine functions through its in-house artificial intelligence platform, AskHR. Internally developed to handle a wide range of support tasks—like payroll processing, vacation requests, and documentation—the system was touted as a cornerstone in IBM’s broader AI transformation strategy.
The move initially appeared to align with a growing trend in the tech industry, as companies from Google to Spotify sought to drive operational efficiency by cutting support staff in favor of machine-led systems. At IBM, the initiative achieved its primary objective: 94% of HR processes were automated, with over 11.5 million interactions handled by AskHR by 2024. Productivity gains reportedly reached $3.5 billion.
But the anticipated decline in workforce size never came. Within a year, IBM’s headcount had grown—not shrunk.
AI Integration Led to Layoffs—And Then New Hiring
As IBM implemented AskHR, it reduced headcount in administrative and back-office roles. These were jobs categorized as highly automatable: process-driven, low in complexity, and repeated across the organization. By all available internal metrics, the transformation succeeded in its efficiency goals.
Yet CEO Arvind Krishna publicly acknowledged that the outcome did not result in a net loss of jobs. In a statement to The Wall Street Journal, Krishna confirmed, “Our total employment has actually gone up, because what [AI] does is it gives you more investment to put into other areas.” Those areas included software engineering, sales, and marketing—roles characterized by high cognitive demand and interpersonal nuance, which AI systems still struggle to replicate.
IBM reinvested the operational savings from AI integration into hiring for roles with strategic value. These were not replacements for laid-off employees but new positions aligned with IBM’s long-term business priorities. The transition supports a reallocation model: automation lowers costs in some functions, freeing capital to scale others.
AI Replaces Tasks, Not Entire Roles
The distinction between tasks and jobs is key. IBM’s example underscores that while discrete responsibilities—like submitting a leave request or updating an employee file—can be automated, the broader roles containing these tasks often include elements requiring human oversight.
AskHR achieved near-total automation of standardized tasks. Yet 6% of queries still required human involvement, often due to ambiguity, contextual complexity, or emotional sensitivity. These edge cases are nontrivial in fields like HR, where trust, clarity, and interpersonal judgment remain essential.
This reality is echoed in other sectors. The article cites Duolingo as a parallel case: the language-learning platform leaned heavily on chatbot technology, only to rehire human staff when automation failed to meet service expectations. These outcomes challenge narratives that frame AI as a seamless, cost-cutting replacement for human labor.
A Hybrid Workforce Emerges
In its current form, IBM represents a hybrid model: AI handles repeatable workflows, while human employees focus on high-value functions. The AskHR platform dramatically improved service outcomes—its Net Promoter Score (NPS) increased from –35 to +74—demonstrating the system’s utility for routine interactions.
Still, the company’s hiring strategy reveals that full automation remains unrealistic. IBM’s growth in non-automatable roles reflects a broader recalibration of labor strategy: use machines for scale and consistency, but rely on people where adaptability and judgment are essential.
This approach isn’t unique to IBM, but the company’s scale makes it a visible test case. With more than 270,000 employees worldwide, IBM’s ability to reconfigure its workforce without reducing overall size may offer a glimpse into how large enterprises will manage automation transitions going forward.
Reshaping the Employment Contract
What IBM demonstrates is not simply a shift in tools, but a fundamental reordering of how work is distributed—and what kinds of work are being valued. Clerical and support positions are increasingly vulnerable, not due to performance or relevance, but because their tasks are now economically more efficient when managed by algorithms.
Yet as those positions disappear, new ones emerge. The net effect is not a reduction in labor, but a redefinition of what labor means inside a technology-led enterprise. IBM’s automation strategy has enabled it to scale differently—not by reducing its human workforce, but by refining it.
The open question now facing both policymakers and business leaders is whether this model is scalable across sectors. IBM had the internal infrastructure, capital, and technical capability to retrain and reallocate. Most companies do not. Without corresponding investments in digital workforce mobility, the automation divide could reinforce structural inequality rather than mitigate it.
