Chatbot in the driver’s seat: tests xAI’s Grok on a Tesla in NYC

On a cloud‑laden afternoon in New York City, a production crew accompanied a Tesla Model Y owner through a series of downtown streets while the driver engaged xAI’s Grok chatbot through the vehicle’s infotainment system, thereby creating a live demonstration of how a large‑language‑model interface can coexist with Tesla’s supervised Full Self‑Driving (FSD) mode, a configuration that nominally promises to keep the driver attentive yet hands‑free.

The sequence began with the driver activating the vehicle’s FSD suite, which subsequently assumed control of steering, acceleration, and braking under a supervisory framework that still required the occupant to monitor the environment, after which the reporter prompted Grok with a variety of conversational queries ranging from navigation assistance to casual commentary, each interaction being displayed on the central touchscreen while the car navigated a pre‑planned route through Manhattan’s congested arteries, thereby allowing observers to witness the simultaneous operation of two distinct artificial‑intelligence systems within a single platform.

Throughout the ride, the driver’s responses to Grok’s output demonstrated a reliance on the chatbot’s ability to deliver context‑aware answers, yet the integration exposed a procedural inconsistency wherein the vehicle’s safety protocols occasionally paused or overrode the chatbot’s suggestions to maintain compliance with traffic regulations, a behavior that highlighted the underlying hierarchy of safety over convenience but also underscored the absence of a seamless coordination layer between conversational AI and autonomous driving software.

By the conclusion of the test, the episode illustrated a predictable tension between the promise of conversational assistants that ostensibly aim to augment driver experience and the regulatory and engineering realities that constrain their deployment in real‑world traffic, a tension that, while not unexpected given the current pace of AI integration, nevertheless points to a systemic gap in industry standards that permits experimental pairings of disparate AI modules without a clear framework for accountability or performance validation.

Published: April 25, 2026