We are obsessed with the removal of friction. Every software update, every new LLM integration, and every automated workflow is sold as a way to "streamline" the messy process of being human at work. We want the result without the labor; we want the insight without the synthesis.
But time is a conserved quantity. It never truly vanishes.
When a developer merges AI-generated code without a line-by-line audit, they have indeed saved their own afternoon. Yet that "saved" time is merely transformed into technical debt, deferred as a future risk, or offloaded as a grueling debugging session for a colleague downstream. Efficiency for the individual frequently manifests as complexity for the collective.
This creates a vacuum where responsibility used to live. In an organization, when it becomes unclear who is actually "authoring" a result, the structure doesn't become more agile—it becomes more fragile. We find ourselves in a panopticon of processes where everyone points to the system, but no one raises their hand.
The central question of our era is: Can we actually "own" the decision of a machine?
For twenty years, the legal and ethical debate over autonomous driving has hit the same wall. When a self-driving car fails, we scramble to find a neck to wring. Is it the "driver" who was told they didn't need to steer? Is it the engineer who wrote the black-box algorithm? Or the vendor who sold the promise of safety? We’ve learned that "explainability" is often a ghost—knowing how a neural network arrived at a disaster doesn't undo the damage.
Passing the buck to the AI vendor is a seductive legal escape, but it’s a moral dead end for the practitioner. If we outsource the "why" to a third-party API, we aren't just outsourcing labor; we are outsourcing our standing as professionals.
To reclaim our sovereignty, we don't necessarily need more control—we need more skin in the game. Real human agency isn't found in the ability to press a "cancel" button on a screen. It’s found in the willingness to be the one who explains the failure when the machine falls silent.
If we allow accountability to be rationalized away so efficiently that "no one is to blame," we haven't achieved progress. We’ve simply negotiated a loss of control in slow motion. The ground truth isn't found in the output; it's found in the person who is willing to sign their name to it.