Jeremy Grantham, founder of a prominent asset management firm, recently signaled his intent to offload tech holdings, comparing the current AI craze to the historical over-investment cycles seen in railways or the early internet. While these technologies eventually became essential utilities, the initial rush failed to deliver the outsized financial returns investors originally projected. This skepticism is increasingly backed by operational data: an S&P Global survey revealed that 42 percent of organizations abandoned their AI initiatives in 2025, a sharp increase from 17 percent the previous year.

The friction is most visible in manufacturing, where the complexity of real-world variables often outpaces algorithmic capability. Ford serves as a cautionary case study; the automaker attempted to streamline development through automation, only to discover that its AI systems lacked the nuance of veteran engineers. The resulting loss of institutional knowledge forced the company to rehire 350 experienced staff members to fix data interpretation gaps. A 2024 RAND report highlights that over 80 percent of industrial AI projects fail, largely due to poor data quality and an inability to account for volatile, non-repeatable environments. As companies grapple with these limitations, the market is beginning to recognize that AI acts best as a tool to augment human judgment rather than a wholesale replacement for it.