The return of craftsmanship in the age of artificial intelligence

Debates over whether artificial intelligence will replace human labor are often framed through the same dichotomy: either the machines will win, or the humans will. Ford's latest quality-focused move, however, disrupts this simplistic opposition. Over the past three years, the company has hired, rehired, or promoted nearly 350 experienced technical experts. These experts are mentoring young engineers, managing design reviews, identifying potential failure points before they reach production, and contributing to the development of AI-based quality tools.

On social media, the event was summarized as, “Artificial intelligence failed, Ford brought back 350 engineers.” It is a striking, yet incomplete sentence. Ford has not abandoned artificial intelligence; it continues to use vision systems, automated testing, and software validation tools in its factories. What has changed is not the investment in technology, but the belief that technology alone would be sufficient. The truth Ford has discovered is this: You cannot guarantee quality solely with more data, cameras, and automation.

Because quality is not just about spotting a defective product in the process or at the end of the line. The real issue is being able to foresee why that error might occur during the design phase itself. An experienced engineer can sense that a part will cause problems during assembly, even if they see nothing against the rules in the technical drawing. They know from past experience that a tolerance might be acceptable on paper but create deviations in mass production, that a software change might unexpectedly conflict with another system, or that a supplier might reproduce the same problem under specific conditions.

Not all of this information is written in procedures. In management science, this is called “tacit knowledge”: the mastery of knowledge formed through experience that cannot be fully put into writing or rules. Over the years, humans learn not only the recorded results from the malfunctions, failed attempts, customer complaints, and production crises they have experienced, but also the context. They distinguish which data is truly important, which measurement might be misleading, and which small sign should be considered a harbinger of a major failure. Artificial intelligence can find patterns in the data presented to it; however, it cannot spontaneously create an institutional memory that has never been transferred to the system.

The statements from Ford executives also point to this. The company admits that in past years, it did not place enough importance on the experience of senior engineers who had gone through numerous product development cycles. Once the company realized this deficiency, it began placing experienced technical experts back at the center of the process. These individuals are not just auditors finding errors; they serve as guides who establish connections between different engineering fields and feed AI systems with more accurate information.

The results are remarkable. Ford rose to the top spot among mass-market brands in J.D. Power’s 2026 U.S. Initial Quality Study. The brand's score improved from 193 problems per 100 vehicles to 152 problems in one year. Thus, Ford became the leader in this category for the first time since 2010. The F-150, Mustang, and Super Duty also ranked first in their respective classes.

The transformation at Ford is not just about rehiring experienced experts. In 2023, the company consolidated its vehicle engineering, manufacturing, supply chain, and quality units under the same industrial system; strengthened weekly design reviews; and involved suppliers in the development process earlier. While factory workers benefit from AI-supported vision systems, vehicle software is tested with hundreds of thousands of automated scenarios before reaching the customer. In other words, the recipe for improvement is not “shutting down machines and calling back humans,” but connecting processes, preventing errors before they are born instead of weeding them out after production, and placing technology under the supervision of expert judgment.

Still, one should not get carried away. The J.D. Power study is based on problems reported by new vehicle owners in the first 90 days and service records, not the entire lifespan or long-term durability of the vehicles. In the overall brand rankings, Porsche is ahead of Ford with 138 problems per 100 vehicles. Genesis is also slightly ahead of Ford’s 152 score with 151 problems. Moreover, Ford’s historical issues regarding recalls have not completely disappeared. Therefore, there is no “miracle” here, but a strong improvement in the right direction.

This distinction is important. Because reading the Ford case as “Humans defeated artificial intelligence” would be just as shallow as the claim that “Artificial intelligence will replace all engineers.” The result the company has achieved is not the product of a race between human and machine, but of the right division of labor. Artificial intelligence can scan thousands of images in seconds, test hundreds of thousands of software scenarios, and catch patterns that the human eye might miss. In contrast, the experienced engineer decides which question needs to be asked, which result is unusual, and what a finding means in real production conditions.

The fundamental problem here is not technology, but the management philosophy. When companies remove experienced employees from the system in the name of cost-cutting or rapid digitalization, they do not just reduce payroll expenses. They also lose the memory of past mistakes, the master-apprentice relationship, and the capacity for judgment that kicks in during moments of crisis. Then, they try to buy this information back from databases, consulting projects, or a new AI application. Yet, some information cannot be brought back once the people who carry it leave the institution.

For this reason, it is necessary to place an “institutional memory balance sheet” alongside AI investments. Which critical information resides only in the minds of a few employees? How is this information transferred to new generations? With whose evaluations are AI systems trained? When technology detects an error, does the human competence to understand the root cause still exist in the institution? Automation performed without asking these questions can produce fragility, not efficiency.

Ford’s experience is important beyond the automotive sector. In many fields, from healthcare to banking, and from media to engineering services, companies see artificial intelligence as a shortcut to reducing headcount. Yet, especially in high-risk and complex jobs, the real gain comes not from removing humans from the system, but from increasing the quality of decision-making. Devaluing an experienced employee and then calling them back to train an AI is not digital transformation; it is an expensive management error.

The lesson Ford learned is actually quite old: Quality is not created by checks made after the product leaves the line, but by the involvement of people with the right knowledge in the process from the very beginning. Artificial intelligence can accelerate, expand, and make this process more visible. But without the human who knows what is important, who is suspicious where the data is silent, and who recognizes the trace of a previously experienced error, the system remains incomplete.

Because quality is not just data. Quality is preserved memory; it is an investment in the future.