Former Tesla data labelers say they would not trust Full Self-Driving

Seven former Tesla data labelers told Reuters they would not trust the company's Full Self-Driving feature to operate a vehicle. The workers had been responsible for labeling data used to train the AI-powered driver-assistance software. One former labeler stated that repeated failures observed during their employment led them to conclude the system could not be relied upon.

Background on data labeling role Data labelers review video and sensor inputs to mark objects, lanes, and traffic signals so the software can learn to recognize them. Their annotations form part of the training data for the Full Self-Driving feature.

The seven individuals spoke on condition of anonymity, according to Reuters. They described instances in which the software did not correctly interpret road conditions during testing. No company response was included in the Reuters report.