You’ve probably noticed the frantic race to turn Jira tickets into automated pull requests. It feels like every other week, some tech leader is on a stage promising that AI will finally solve our engineering productivity crisis. But let’s be brutally honest: We’re using billion-dollar algorithms to do the work of a five-minute uncomfortable conversation.
The seductive ease of deploying AI is undeniable. You swipe a corporate credit card, provision an enterprise license, and suddenly you can tell your board that you’ve ‘leveraged machine learning to accelerate delivery.’ It’s a technological escape hatch. Because the alternative—the messy, slow, excruciating work of fixing a broken engineering culture—requires something a piece of software can never provide: accountability.
Let’s talk about what’s actually happening on the ground. We have junior devs relying on what is essentially automated StackOverflow copypasta on steroids, amplifying all the downsides of inexperience. We have principal engineers building elaborate, demotivating systems just to avoid mentoring humans. And we have C-level managers who create environments where good ideas go to die. AI isn’t a productivity multiplier; it’s a spotlight on your existing dysfunction.
The DevOps movement tried to warn us years ago that culture eats process for breakfast. And what did we do? We ignored the culture part, grabbed the tooling, and kept right on abusing our engineers. Now, history is repeating itself with AI. It is infinitely easier to deploy a large language model than it is to look a toxic manager in the eye and say, ‘You are the bottleneck.’
Here is the uncomfortable truth every leader needs to hear: You can’t automate your way out of a culture you were too cowardly to build. Good culture is the ultimate, un-replicable productivity hack. It’s the psychological safety to fail, the trust to move fast, and the shared purpose that makes people actually want to give a damn. No LLM can generate that.
When you deploy AI into a toxic system, you don’t get faster output; you just get demotivated output, faster. You automate the misery. The cultural debt you’ve been ignoring will eventually cap any technical upside your shiny new tools promise. If your organization is fundamentally broken, AI won’t fix it—it will just help you produce broken things at scale.
So, put the AI deployment on hold for a second. Ask yourself the hard question: Are we using this tool to multiply the brilliance of a healthy team, or to mask the rot of a broken one? If it’s the latter, no amount of compute power will save you. Stop looking for a technological savior. Start doing the human work.
FAQ
Q: Isn't AI actually making developers faster?
A: It’s making them type faster, not ship better products. If your system penalizes risk and rewards busywork, AI just helps you generate garbage at scale.
Q: What's the practical implication for leadership?
A: Before buying an enterprise AI license, audit your management chain. Fire the toxic managers first. The ROI on fixing culture is infinitely higher than any tool.
Q: What's the contrarian take?
A: AI adoption is often a symptom of leadership failure. If your first instinct to fix productivity is buying software rather than talking to your team, you've already lost.