I try to be brutally honest in my writings, warts and all. No one makes more mistakes than me, and hopefully others can learn from my shortcomings. This is the most painful newsletter I’ve written, and the one I struggled the most with since some of my covid pieces.
Two years ago, I told my teams that no one would lose their jobs because of AI.
I was wrong. A few weeks ago, I laid off 25% of the team.
I’ve been the biggest AI fanboy since day 1, but I still underestimated the impact it would have on our businesses. I am embarrassed by how badly I missed the mark. I’m still not sure I should be writing about this.
I took a lot of pride in zero layoffs during covid. I figured if we hunkered down, we could ride it out together.
As business slowed during the brutal rate tightening cycle, I was reluctant to let go of anyone that we had poured so much energy into training. The labor shortage from covid was a fresh lesson. I didn’t want to regret losing talented people and then play catch up.
But this was different. Over the last few months, more and more processes were automated and perfected. This wasn’t a storm to be weathered. There was no hunkering down. No light at the end of the tunnel. Suddenly, I was asking uncomfortable questions.
“Well…what does that person do now?”
It’s hard to overstate how quickly this happened. In less than 30 days I went from “AI will never replace anyone here” to “that role doesn’t exist anymore.” The pace of change was staggering.
There was no universe where that role would ever exist again. It was never coming back. We were never going back.
It felt…inevitable.
These are great people. Great teammates. They’ve done what we asked. I looked for ways to repurpose them into other roles, but those roles have also been impacted. A huge chunk of daily work has been automated across the platform, so there wasn’t a great place to put them.
One of the best loan abstractors had been with us for seven years. I hugged her while she sobbed, telling her, “This isn’t your fault. This isn’t your fault.” Nothing in my 25+ years of working had prepared me for this moment. She’s probably furious with me. I don’t blame her. They all have a right to be mad at me. I let them down.
But I also feel like it was the right decision for the other teammates. Our customers. The companies. I didn’t sleep for weeks leading up to this decision because of the impact on them and their families, but not because I was unsure. Once I concluded it was inevitable, I saw no point in dragging it out.
I’m old enough to remember automation in the car industry in the 1980s. Factory workers were being replaced by robots, spearheaded by the Japanese manufacturers. I found an article from 1984 that said one robot replaced 1.4 workers, impacting 1,044 employees at that plant alone. In the moment, it was an incredibly painful transition. But in the long run, it led to much better cars and more (but different) jobs.
Toyota has a philosophy of jidoka - automation with a human touch. This is where teammates add value now, not in the execution of repetitive tasks.
The judgement. The insight. The battle scars from trial and error in the real world, not from scraping the internet for examples.
The wisdom.
The team is morphing into a smaller group of experts. They use their judgement to detect issues. Use that wisdom to oversee the assembly line and jump in when needed. Change the AI process. Build new ones. Use their wisdom to be better for our customers.
Today’s writing comes as our youngest graduates high school today. What will his job prospects be after college? How does he gain wisdom if owners like me stop hiring? Does that create a vacuum in institutional knowledge that bites me 10 years from now?
In the long run, I believe AI will be hugely net positive for the economy and the labor market. But in the near term, I continue to believe there will be a painful dislocation. I think many will drag it out longer than necessary, but ultimately arrive at the same place.
Then again, I was wrong about my own teammates being replaced by AI. I hope I’m wrong about this, too.