IBM pauses hiring for roles it expects AI to replace
The estimate covered non-customer-facing roles over five years; IBM later said the actual toll was a couple of hundred HR positions, offset by hiring elsewhere.
- Culture & impact
- Money & business
- Notable
IBM chief executive Arvind Krishna told Bloomberg that the company would pause or slow hiring for roles in “back-office” functions such as human resources, which he judged the most exposed to being automated by AI. He estimated that around 30% of the roughly 26,000 employees in those non-customer-facing functions — about 7,800 jobs — could be replaced by AI and automation over the following five years, and said IBM would simply not backfill many such roles as people left rather than announcing layoffs directly.
Krishna framed the change as gradual attrition rather than a redundancy programme: IBM would keep hiring for software development, sales and other customer-facing work, and the pause applied specifically to internal administrative functions where the company judged AI tools could increasingly handle routine tasks such as processing employment verification letters or answering standard HR queries.
The figure of “7,800 jobs” was widely reported as a concrete AI-driven layoff number, though it was always a projection over five years rather than an announced cut, and IBM’s own later account undercut the scale of the framing. By 2024, the company said its AskHR assistant handled the large majority of routine HR questions automatically, but that the roughly 6% of queries still requiring human judgment had created service gaps; Krishna subsequently said AI had actually displaced only “a couple hundred” HR roles, with savings redirected into hiring for software engineering, sales and marketing, and IBM’s overall headcount grew rather than shrank.
The episode became a frequently cited, and frequently overstated, early example of “AI is taking jobs” headlines — an announcement of hiring caution across two years turned, in retelling, into a single mass layoff, while IBM’s own subsequent numbers suggested displacement at a much smaller scale than the initial estimate implied.