Hi everybody, and welcome to episode number 521 of For Immediate Release. I’m Shel Holtz

And I’m Neville Hobson. Here’s a story that should make every one of us pause before we get too comfortable handing things over to AI. Ford, the automaker, has just rehired somewhere between three hundred and three hundred and fifty veteran engineers. Note the word rehired. The company had let them go in recent years as it leaned into AI-driven quality checks. Ford calls them greybeard engineers. That’s not a throwaway nickname. It’s the whole point of the story. These are the people with decades of experience across multiple product cycles, and Ford let a lot of them go only to discover it needed them back because the AI wasn’t working the way Ford expected. We’ll look into what happened right after this  Charles Poon, Ford’s vice president of vehicle hardware engineering, put it plainly on a call with reporters. Here’s what he said: “Mistakenly, we thought that by just introducing artificial intelligence and ingesting the design requirements that we had, that would produce a high-quality product.” Think about that for a moment. Ford didn’t skip a step. They fed the AI everything that was written down, every design requirement, every documented specification. It still wasn’t enough. And it wasn’t just one system. Ford had installed around nine hundred AI-assisted cameras on the production line specifically to catch quality issues. Nine hundred cameras, and still they couldn’t replace the trained eye of an experienced technician who knows what a problem looks like before it becomes a visible defect. Ford’s chief operating officer, Kumar Galhotra, added more context. He said the company had been leaning more and more on automated quality systems, and the results were disappointing. Teams across software, hardware, manufacturing, and supply chain had also been working in isolation from each other, which meant defects were being caught late and fixed under pressure rather than prevented early. Galhotra described this as a find and fix mentality that Ford is now trying to move away from towards genuinely preventing problems before they start. The returning engineers sit right at the center of that shift. They now run mandatory weekly quality and design reviews, hunting for failure points before a single part reaches the factory floor And here’s the part I think matters most for us. A lot of the people who held that hard-won judgment had already walked out the door to suppliers, to retirement before anyone at Ford thought to capture what was in their heads. Poon admitted as much. “Over prior years, we didn’t pay as much attention as we should have to the experience of our most knowledgeable engineers who have been with us through many product cycles,” he said. So Ford had to buy that expertise back three years into this process at real cost. Was it worth it? By Ford’s own numbers, yes. The company has just topped the J.D. Power Initial Quality Survey for mainstream brands for the first time since twenty-ten. That’s sixteen years. CEO Jim Farley says the rehired engineers are already contributing what he called literally hundreds and hundreds of millions of dollars in savings, largely through reduced warranty and recall costs. Ford’s even projecting around a billion dollars in cost reduction this year on the back of this quality push. Now, here’s a tension worth sitting with. This is the same Jim Farley who said publicly on other occasions that AI is gonna replace rough-roughly half of all white-collar jobs. And yet here’s his own vice president standing in front of journalists explaining that Ford’s entire quality turnaround depended on bringing back the very human expertise the company thought it could do without. To be clear, this isn’t really a story about AI failing and humans winning. Ford isn’t walking away from AI. Those returning engineers aren’t just doing inspections. They’re training junior staff, and they’re reprogramming the AI tools themselves, feeding them the judgment that design requirements alone couldn’t capture. It’s a hybrid fix, not a retreat. But for anyone in our line of work, comms, knowledge management, anyone thinking about where AI fits into institutional expertise, there’s a sharp lesson underneath all of this. Documentation isn’t the same as judgment, and once the people carrying that judgment are gone, you don’t get it back for free or quickly or easily. I think there’s a bigger question here, too, about how organizations are handling this handover between human expertise and automation, and whether Ford’s experience is a one-off or a warning sign for a lot more companies than just car manufacturers. Shel

Yeah, I think it is a warning sign. But I, I don’t think it’s a trend necessarily, the idea that AI AI layoffs are being reversed everywhere. Yeah I’d just be careful about overstating that. There’s really only a handful of well-documented company examples of this. I think it’s easier to say that the way many organizations overestimated how quickly AI could substitute for the experienced judgment of their staff, I think that’s a reasonable way to look at it. Increasing number are recalibrating toward AI human collaboration rather than there are examples though. IBM has reversed course. They didn’t rehire the same people, but they’ve really reversed course on this whole replacement idea. An AI system deployed to take over HR work handled about 94% of incoming requests, but the 6% it couldn’t resolve including situations in- involving ethical judgment, really revealed the limits of all of this into the hands of a large language model. And then the company announced that it planned to triple its US entry-level hiring this year. That’s a pretty significant reversal. Klarna’s the one that … that’s the poster child for all of this. They were one of the first to announce that they were going to replace their customer service with AI. A year later their CEO publicly reversed course, admitting that customer experience had gone down the tank, quality had fallen the company had over-prioritized cost savings which most companies seem to be doing. They’re looking at cost savings and not other ways AI could really improve things or even help the organization grow and that human customer service remained essential. So they went back to hiring customer service representatives, and they expanded their human support. They la- even reassigned engineers and marketers into customer support roles while they were busy rebuilding the support organization that they had decimated. CEO, I think it was he who was quoted saying, “Cost, unfortunately, seems to have been too predominant evaluation factor.” that said I think it is worth noting that according to one research organization, I hadn’t heard of them before, but OrgView, 39% of business leaders made employees redundant due to AI deployment, and among that number, 55% admit that wrong decisions about those redundancies made, 32% of US hiring managers said they eliminated a role primarily due to AI later rehired for the same or similar positions, that according to Robert Half. This is definitely something we need to be looking at. I think what Ford has done is, as you say a warning sign, but I don’t think there’s a clear trend yet that people who off in order to accommodate AI are suddenly reversing and rehiring yet

No, I agree. That doesn’t seem to be a trend. What is a trend is the laying off element of it as opposed to rehiring. So the… I think there, there’s a good question for our audience in all of this. How many organizations right now are automating roles without first extracting what the people they’re letting go know, their knowledge? Ford’s mistake wasn’t using AI, I mentioned earlier. It was letting the knowledge holders leave before capturing anything from them. That’s a sequencing failure, not a technology failure. So is knowledge capture before AI rollout ever ac-actually built into transformation plans, I wonder? Or is it always an afterthought that only gets addressed once something breaks?

It’s an interesting question.

Yeah.

when we first started this show 21 years ago, knowledge management was a big issue, and we talked about knowledge management and knowledge management systems, and they have fallen by the wayside even though there are still efforts to build them out. I think the need for clean, consistent data in order to run AI inside your organization is bringing the idea back, although not with the kind of notoriety that it had. There were books being written about knowledge management back 20, 20 years ago. But knowledge capture is an interesting thing because, there’s so much ta- tacit knowledge walks out the door and goes home every night. It’s not implicit knowledge that you can store in a database, and this was the problem with knowledge management systems, right? Is you’re trying to capture tacit knowledge, and it had to be tagged, which meant the people looking for it had to know which tag was used in order to retrieve it. That knowledge could be sitting there, but if you’re searching with the wrong tag, you’ll never find it. So it, it remains a challenge how to do this. I remember one effort at knowledge capture. Was it Intel? I’m gonna struggle to remember, but it was an early wiki where people would just leave their knowledge that occurred to them. “Oh, this is how you do this,” and they were trying to capture it that way. I don’t know if that’s still around. I never hear about it anymore. I- like I expect, I can’t even remember for sure what company it was, but have been efforts at this.

Yeah

and with the AI situation and companies like Ford finding the knowledge isn’t there once the people leave and the AI isn’t equipped to handle everything just based on was in the databases that it had been fed, I think we’re gonna have to revisit this whole idea of knowledge management and get it right this time because it’s still far from perfect

Yeah. Thinking about Jim Farley and what he said that AI will replace half of white-collar work, while his own VP describes a turnaround built on rehiring the humans they’d let go. I-is this Ford being inconsistent, or is it actually a realistic picture, AI displacing some roles while creating new dependency on a smaller pool of s- pool of senior experts? I wonder if that’s it. There’s also a related point to that. Could this be a warning? You, y-you mentioned– You used that word a minute ago in this case about h-hollowing out effect. If companies keep pushing out mid-career and senior staff in favor of AI, who trains the next generation of greybeards when today’s greybeards retire for good? ‘Cause that will happen sooner or later.

There’s also this idea of institutional knowledge that is not necessarily captured in any systems. It’s because they have been there, as was pointed out in this article, through multiple product cycles. There are things that they learned that are not a specific quality measure or whatever it is that they’re using in their work that now is gonna be done by the AI. They bring that institutional knowledge to the job, and y- you may not need as many of them as you had before if the AI can legitimately do some of this work. But some of those people who have been there through those product cycles and have acquired that institutional knowledge, they’re still gonna need to be there. I don’t know how you capture institutional knowledge. I have seen this in an organization that shall remain unnamed, but that you and I are b- both very familiar with, that, that hollowed out staff and let its institutional knowledge go, and is paying a dear price for that

Yeah. It got me thinking a bit about I guess drawing upon some science fiction movies that I’ve seen over the years but also now looking at where we are with technology tools that make it very easy for people to literally dump their knowledge about something into a device or a system that records that saves it. Could we be seeing a, the beginnings of something or the idea certainly that may catch on with people that this needs to be built into behaviors in an organization that depending on your role, and I can’t say it could be everyone, but maybe it should be, that you’ve got some means that you work on a project and part of your in a sense, your the kind of conditions of your employment, let’s say, is that on completing a project or planning a project and then com- wha- whatever it might be, you’ve gotta record your thoughts on the planning and execution of that project how you did this or that. And that’s then saved. Okay. I haven’t– I’m not even getting into privacy i-issues or data protection or none of that. That may well be the case. I’m thinking of one science fiction film I saw where, a few, some years ago now, where people had, It reminds me of Plaud, actually, a little credit card sized tool.

say

Yeah.

that. Yep

That they spoke into, recorded something. It then saved it to their account, let’s say, on their employer’s website or someplace or whatever it might be, that then broke it down into all the constituent elements that the company needed to have in order to retrieve that information on demand or make use of it in some other way. So some of the tech exists to do that, but certainly not on an organized scale such as what we’re discussing that might be. But how long might it be before that happens?

Aside from the privacy concerns, wearing a device that captures everything you say, having all of that from all employees fed to an AI that’s able to sort it out and put it in the appropriate place so that now that becomes part of its knowledge base, I can see that. I just can’t see employees tolerating it very well. I certainly wouldn’t want to have every word I’m saying during the day recorded and saved by the organization, and I’m a senior executive. That’s just ethically wrong I think. But technically I think we’re not that far away from

No, we’re not. We’re not.

do that now

You raise a good point but I would counter that in a sense arguing that it’s gonna be if you don’t do it, someone else is gonna do it, and you’re out of a job if you don’t do that. I’m thinking of some of the things that you see around you nowadays as a matter of course, ranging from things like police, ambulance workers fire department people wearing video camera, body cams that record everything. I have a friend of mine who’s just bought one that he feels safe wearing this on his commute to work on the train. He goes to London and then on the underground train. I can, I can–

use case. These are, these f- yeah I think,

it’s not. It’s not. It’s rec- it records the picture of him. That could be– The purposes might be very different, but the tech would be the same.

Oh, the text’s the same. The

Yeah.

is

Yeah. For now it is. Yeah, it is. But this could become a matter of course. I’ve considered that too. If I were still actively working in a traditional job, getting in a commute to London or whatever it might be, I would probably get a body cam

Yeah, the okay, the use case there is voluntary, right? You choose

I do have that choice

Think of Meta, which was capturing the keystrokes of every employee in order to train its models, and the opposis- opposition to that grew so loud that they recently said, “Okay, this is now a voluntary program. Only employees who opt in will have their keystrokes recorded.” And it– now you’re talking about every

Yeah, I know, but,

the workplace?

sure.

gonna happen. Yeah.

so this is– enters that bigger debate precisely on is it gonna happen or is it not? So in the case of Meta, what they did, I’m thinking back now to the Cambridge Analytica scandal of twenty eighteen, they did this ov- covertly without telling people, and then they lied about it all. So why the hell would you work for a company like that in the first place? That, crosses, crosses my mind. But

Yeah

I think the one other point occurs to me that the in the case of Ford, that has got me thinking. The fix they had, i.e., rehiring the people they, they got rid of, works because those engineers were still available to rehire, right? So what happens to the next company that has the same realization five or ten years from now when the generation with that tacit knowledge has genuinely retired and isn’t coming back? That’s arguably the real warning sign, not this instance, but the next one

Oh, absolutely. They’re gonna have to go for the next best thing, which is hiring humans that can handle the more complex problems that arise, but they don’t have the institutional knowledge. They haven’t been through any product life cycles. Maybe they hire away from competitors. Maybe they hire their old engineers back by offering them more money than they ever dreamed of, but yeah, that sort of defeats the purpose too, doesn’t it? No, I think what has to happen is that the companies that are considering replacing people with AI don’t just count the number of who are doing the job and how much an AI can conceivably do, and then get rid of the equivalent number of staffers. You’ve gotta be really strategic about this. I think you need to inventory all of the tasks that you’re talking about performing and identify which of those are better done by AI. Heaven knows there are things that AI is not good at,

Yeah

pretty widely recognized sets of activities. So looking at what you’re gonna need people for and then determining, okay, if we’re gonna let people go how many we still need in order to avoid the kind of outcome that Ford experienced? And I think if they’re strategic, if they apply some formulas to this again, inventorying absolutely every task you’re gonna be looking at replacing and figuring out which of those are going to work with AI you’re probably gonna end up making better decisions and not have to do what Ford did. The other thing that I would look at though is how does all of this align with your values? Because no matter what you do, it is in conflict with your stated corporate values, which are hanging on the wall in the conference rooms and, employees have copies of them, and in my company, they’re on every truck that’s out there you’re gonna end up with a very disengaged cynical workforce

Yeah. The cynical part of me suggests that’s actually happening a lot already. G-question mark, do they really believe those values on the side of the trucks down the road, all that kind of stuff? So it’s a kind of fluid environment we seem to be in. And I’m wondering w- as well, will we begin to see, I don’t know on your CV or your resume on LinkedIn one of your major selling points is gonna be the fact that you have deep institutional knowledge. You probably wouldn’t wanna say about your employer, but what you really want to be saying is about the industry or about something that sa- that kinda says, “I’m very portable from the get-go.” So maybe we’re gonna see that

Oh, could be. It would certainly be better than open to work as a little arc on the pr-profile photo on LinkedIn. And that’ll be a 30 for this episode of “For Immediate Release.”