Former Tesla labelers reveal FSD failures endangering pedestrians and animals

Former Tesla labelers reveal FSD failures endangering pedestrians and animals Former Tesla labelers describe raw dashcam footage showing animals hit and near misses with children, raising urgent questions about FSD safety and company claims. Based on data from Reuters In a Utah office, hundreds of Tesla employees are carefully reviewing video footage gathered by cars using their Full Self-Driving (FSD) feature. Some clips show the cars hitting cats, dogs, or deer, as well as other, more common crash scenarios. Sometimes they don’t brake before a collision. The cars frequently accelerate. And occasionally workers see near-collisions with children playing in the street. They are known as “data labelers” – they train Tesla’s driver-assistance AI system. They label examples of successful and unsuccessful driving behavior and pass issues on to engineers working on improving the system. Tesla CEO Elon Musk says that FSD will soon make all of the company’s cars fully autonomous. Yet interviews with nine former data labelers and a former autonomous driving engineer reveal that the technology has continued to struggle with basic maneuvers in recent months – such as avoiding emergency vehicles or stopping for school buses that are loading or unloading students. Despite such dangerous shortcomings, Musk and other executives are increasingly pressing for the safety of FSD, while promoting public demonstrations of full autonomy for investors. Among these demonstrations is the robotaxi pilot in Austin, Texas, launched last June with partial human supervision in the cars and remote monitoring by an operator. Inside Tesla, as these events drew closer, workers put in long hours mapping routes and training software to respond to specific hazards in order to make the self-driving technology seem more “workable” to the public, according to former employees. They noted that such labor-intensive measures cannot be scaled to a broad rollout. These efforts,