Ford had to re-hire former engineers to fix the defects in its machines

Celebrating its new position as Number 1 in JD Power’s first ranking among automakers, Ford is opening up about the challenges it has faced in recent years, especially in relying on automation in design and manufacturing. It turned out that the automation wasn’t as robust as previously thought, requiring Ford to hire more experts — sometimes bringing back former employees — to fix the mistakes made by the company’s robots.

In Ford’s view, AI is powerful and vulnerable. Its effectiveness depends entirely on the type of data used to train the AI ​​models. In addition, the automaker reduced the need for business knowledge acquired by its veteran engineers who had worked around multiple vehicles. And the combination of these events led to the downfall of Ford cars.

“Wrongly, we thought that by just introducing the artificial intelligence and changing the requirements that we had, that we could make something of value,” said Charles Poon, VP of engineering for automotive components, in a briefing this week with the press.

According to Poon, some of the company’s most experienced employees left before all of their acquisitions were transferred into Ford’s manufacturing machinery. This required the bringing back of some of the employees to restructure the system, or in some cases, the young consultants who are currently struggling to maintain the Ford brand. Poon said Ford has hired, promoted, or brought back more than 350 engineers to rebuild this technology. In addition to leading the engineering team, he was also tasked with developing the data collection and AI training that powers Ford’s automotive systems.

“That’s where our experienced engineers encountered problems and identified these problems before they entered the system,” Poon said.

Ford right now leads the industry in the amount of memoryand its good ratings slipped away in the past few years. These problems became more apparent recently, with the problems associated with the launch of the Explorer and the Aviator, the gradual disruption of the covid epidemic, and the apparent growth of the number of its vehicles recalled.

According to Ford’s COO Kumar Galhotra, the automaker finally decided that its strategic direction was too fragmented. Different departments work in silos, and the company has relied heavily on a “find and fix” philosophy that focuses on identifying errors when they appear and fixing them quickly. Although this method may solve the immediate problems, it does not prevent the problems from happening.

“We’re moving away from a catch-and-fix mentality to prevent problems before they happen,” Galhotra said. “We focus on indicators and early signs and outcomes. Stop admiring the problem and start solving it.”

Innovation extends beyond automotive components. Software and digital teams are now working closely with automotive engineering, manufacturing, and sales teams, executives said. And Ford is now trying to combine the speed and flexibility associated with software development with the complexities and requirements of automotive engineering.

Historically, this has not always been the case. Ford only discovered the software’s problems late because it didn’t support the available controls, Poon said. That said, the automaker couldn’t roll out software updates as quickly as electronics companies with the mindset that they “can move quickly and fix it later,” Poon said. Cars, unlike mobile phones, operate in a very secure environment where customers rely on the software working properly from the moment the car is delivered. To fix this, Ford created a dedicated team of 40 people responsible for preventing problems before they happen.

But don’t think that Ford hasn’t committed to integrating AI into many of its systems. The machine maker says it will significantly expand its testing capabilities, adding more than 100,000 AI-powered tests designed to detect side-cases and stress programs in a variety of situations. Because the test mode is automated, software updates can be made quickly and even late, ensuring that updates do not introduce new bugs.

“Because these tests are automated, even if we make late changes to the software, we can re-run the entire process to verify that it works properly before it goes to the customer,” Poon said. “We have established the reliability of the software as its strict performance and strict metrics.”