Uber automated test vehicle kills a pedestrian in Tempe

On March 18, 2018, an Uber developmental automated test vehicle struck and killed a pedestrian in Tempe, Arizona. The NTSB report describes a system that detected her but did not correctly classify her or predict her path. Emergency braking for collision mitigation was precluded by design, with intervention left to the operator.^[1]

What the investigation establishes

The NTSB identifies the distracted operator’s failure to monitor as probable cause. It also identifies inadequate risk assessment, ineffective operator oversight and insufficient responses to automation complacency, associated with Uber ATG’s inadequate safety culture. Pedestrian impairment and insufficient state oversight also contributed (§3.2). A single model error is therefore an incomplete account.^[1]

Alignment relevance and limits

This case makes the human-control question concrete: what must an operator know, notice and be able to change? It does not establish a hidden learned objective or Deceptive Alignment. Compare the institutional response to Tay public chatbot incident in the incidents guide.

Review scope: executive summary and probable cause in the complete original report, adopted November 19, 2019. The crash date and report date are distinct. Raw data, the full docket, legal responsibility and later remedies were not audited.