xAI Grok In-Car Test: Using It While Driving Tesla Full Self-Driving in New York City

Recently, a CNBC reporter joined Tesla Model Y owner Mike Nelson for a real-world test in New York City, experiencing the interaction between xAI's Grok chatbot and Tesla's Full Self-Driving system. Nelson, a lawyer with an auto insurance background, has been using Grok in his vehicle for several months. During the drive, he kept Tesla's Full Self-Driving system active and issued several voice commands to Grok, asking, for example, whether it could adjust the seats or the air conditioning. The CNBC reporter noted that while crossing the George Washington Bridge, the busiest bridge in the U.S., Nelson admitted: “I’m not paying attention to driving at all.” This test vividly highlighted the convenience of in-car AI voice assistants alongside the potential danger of distraction.

Integrated into Tesla vehicles, Grok can handle navigation, answer questions, and more, enhancing the driving experience to a certain degree. However, experts warn that such AI chatbots can easily divert a driver’s attention, especially when engaging the Full Self-Driving system, which still requires active driver supervision. The CNBC on-road experience confirmed this: although Grok can explain the self-driving decision-making process, over-reliance on AI may lead to “over-trust,” causing drivers to neglect road monitoring. In the heavy traffic of Manhattan, this tension between technological capability and safety assurance becomes especially pronounced.

The key to future in-car AI lies in designing context-aware interaction models. For example, the system should automatically detect whether Full Self-Driving is active and, based on that, limit the depth or duration of conversations rather than responding without restriction. The Grok New York test reveals a path toward solving the “black box problem”—letting AI explain its own decisions—but also exposes a new failure mode in safety. The tech industry must enhance functionality while building more robust driver attention monitoring mechanisms, ensuring that convenience does not become a safety hazard. This test serves as a wake-up call for human-machine interaction design in smart cockpits: convenience and safety must evolve together.