Tesla FSD: The Deadly Trap of ‘Almost Perfect’ Tech

Rafi Krikorian, former head of Uber's self-driving division, recounts crashing his Tesla Model X while using Full Self-Driving (FSD) mode on familiar Bay Area roads, totaling the car with his kids aboard but causing only minor injuries. Host John Johnston uses this Atlantic article to explore the episode's core thesis: "almost perfect" tech like Tesla FSD—reliable 99% or more of the time—fosters complacency, eroding vigilance just when rare failures demand quick human intervention.

Krikorian's Crash and Expertise

Krikorian describes a routine drive to his son's Boy Scouts meeting. The Tesla handled the turn until the steering jerked erratically and decelerated unexpectedly; he grabbed the wheel but hit a concrete wall seconds later. Despite his background—running Uber's autonomous program with no injuries in pilots—he was lulled by FSD's flawless prior performance on highways and local roads. He started with highway use, where lane markers aid predictability, but habituated to city streets, shifting from active steering to passive monitoring. No warning signaled the system's confusion, and logs later showed his wheel input without capturing the mental lag in re-engaging.