Two stories about AI and work are circulating this week, and at first glance they point in opposite directions. In one, Ford has quietly rehired around 350 veteran engineers after admitting its AI-powered quality systems could not do the job alone. In the other, the entry-level market for young graduates is the worst it has been in nearly four decades. Read together, they describe the same machine doing two different things to the career ladder: chewing at the top and sawing off the bottom.
Start with Ford, because it is the more reassuring half. According to a Bloomberg report picked up by Tom's Guide and ABP Live, the carmaker went all-in on AI quality control, including 900 AI-powered cameras meant to catch defects on the line. The results disappointed. "Mistakenly, we thought that by just introducing artificial intelligence and ingesting the design requirements that we had, that would produce a high-quality product," vice-president Charles Poon told Bloomberg. The fix was human: bring back the "gray beard" engineers, the people whose decades of hard-won judgment had never been written down, and let them train both the juniors and the AI itself.
Ford is not alone, and that is the point. Commentators have started calling this the "AI boomerang." Forrester predicted roughly half of AI-related layoffs would eventually be reversed. A Robert Half survey found about 29 percent of companies had rehired for the exact roles they had cut, with finance leading at 44 percent. Of 2,000 hiring managers, 40 percent said AI could not replace institutional knowledge and 38 percent admitted they had underestimated the need for human quality control. The lesson keeps repeating: AI is a task machine, very good at slices of a job, and surprisingly poor at carrying a whole one across the finish line.
Now the worrying half. If experienced judgment is what AI cannot replace, where does experienced judgment come from? It comes from people grinding through the unglamorous entry-level work that AI is now very good at: the spreadsheets, the first-draft code, the document review. Fortune, drawing on a Wolters Kluwer analysis, notes that AI succeeds on individual professional tasks 50 to 60 percent of the time but completes an end-to-end project only about 2 percent of the time. So firms keep their senior people and stop hiring juniors. Entry-level roles in professional services have dropped 29 percent since January 2024, and a Stanford study found workers aged 22 to 25 in highly AI-exposed jobs saw employment fall 13 percent.
PwC has a name for the result: "seniorization." Its analysis of more than a billion job postings found entry-level roles in exposed occupations are now seven times more likely to demand skills, like strategic judgment and stakeholder management, that used to appear much later in a career. Employers want apprentices who arrive already expert, which is a contradiction. As one industry essay put it, companies risk discovering years from now that they quietly stopped training their own replacements. The open question is not whether AI destroys jobs in bulk; the data says it has not. It is subtler and slower: if the bottom rungs vanish, who climbs to the top, and who trains the machines when this generation of gray beards retires?