Ford thought it could replace institutional engineering knowledge with AI and the ingestion of design requirements. After quality suffered, the automaker rehired roughly 300 veteran engineers as auditors of every new design and topped the J.D.
Ford publicly admitted it had over-rotated on artificial intelligence in its vehicle design process and is now relying on the institutional knowledge of veteran engineers to recover.
Charles Poon, Ford's vice president of vehicle hardware engineering, told reporters that the company "mistakenly thought that by just introducing artificial intelligence and ingesting the design requirements that we had, that would produce a high quality product" (Bloomberg, 25 June 2026). AI, he said, is "a fantastic tool" but only as good as the information used to train it.
The corrective has been a major rehiring. Ford has brought back roughly 300 veteran engineers, many of them retirees or former employees, and turned them into internal auditors who run mandatory weekly design reviews on every new vehicle program before it reaches the factory floor (The Verge; BBC). The reviews are intended to catch failure points that an AI system fed only design specifications might miss: the kinds of flaws an engineer recognizes only from decades of watching cars fail in the field.
The measurable payoff arrived in late June. Ford topped the J.D. Power 2026 Initial Quality Study, a widely watched U.S. survey of problems in the first 90 days of new-vehicle ownership, for the first time since 2010 (The Verge; Ford corporate statement). The F-150, Mustang, and Super Duty each led their segments for a second straight year, and seven of Ford's top ten models finished in the top three of their categories.
Ford's reversal follows a period when the company had trimmed engineering headcount in parts of its organization tied to AI-augmented design work. Poon acknowledged that "we didn't pay as much attention as we should have" to the engineers who carried that institutional craft knowledge (Bloomberg). Reporting from Fox 2 Detroit and TechCrunch cites rehiring figures described as "several hundred," slightly higher than Ford's own roughly 300 count. The narrow gap reflects different definitions of who counts as rehired rather than a contradiction in the underlying move.
The mechanism is the more interesting story than the head-count number. Ford treated the institutional knowledge embedded in its most experienced design engineers as something that could be replaced by training a system on the company's existing design requirements. That knowledge surfaces only in the judgment that comes from watching parts fail across many vehicle generations, and it rarely makes it into the documents the AI is trained on. The experience that lets a reviewer spot a likely failure before tooling up is not in the specifications; it sits in the people who have been through enough programs to recognize when a design is about to break.
The publicly reported figures do not support claims about AI specifically causing quality defects on its own. Ford positions the issue as having under-invested in the human layer around the AI, rather than as having used AI incorrectly in isolation (Business Insider; Fortune). The trade-off the company is now naming is one of substituting data ingestion for hands-on craft knowledge, not of abandoning AI tools.
The watch item is durability. J.D. Power's initial-quality study measures problems in the first 90 days of ownership, so the 2026 result largely reflects vehicles designed before the rehiring was in place. Whether the auditor-led design reviews produce the same quality gains on programs designed with the expanded human review from the start is the open question for the 2027 and 2028 surveys, and the next read on whether the corrective has actually taken hold.