GM wants to put ChatGPT in your car

Picture this: you’re cruising down the highway, and an unfamiliar warning light flashes on your dashboard. Most drivers would either panic or wait for a red light to frantically Google the symbol on their phone.

General Motors wants your car to handle that conversation instead. The automaker behind Chevrolet, Cadillac, Buick, and GMC is developing an in-car virtual assistant powered by the artificial intelligence models behind ChatGPT.

More mechanic, less chatbot

Scott Miller, GM’s vice president of software-defined vehicles, confirmed the project, saying the new assistant will go far beyond basic voice commands like turning up the heat or skipping a song.

Say you get a flat tire. Instead of flipping through a 400-page glovebox manual, you can ask the car what to do. The system could walk you through the process and pull up an instructional video right on your dashboard screen.

If a diagnostic light pops up, the assistant can explain what went wrong, advise whether you should pull over immediately or if it is safe to drive home, and even book an appointment at a nearby repair shop.

The system runs on Microsoft’s Azure cloud service, drawing on GM’s existing two-year partnership with Microsoft.

Crucially, GM isn’t just slapping standard ChatGPT into the dashboard. Because raw large language models are known for unpredictable answers, GM is building a proprietary, car-specific layer on top of OpenAI’s tech to keep the assistant accurate and grounded in real vehicle data.

Squeezing the smartphone out

For automakers, the real enemy on the dashboard isn't bad software—it's your phone.

Apple CarPlay and Android Auto have effectively turned in-car screens into dumb monitors for smartphones. Automakers dread becoming low-margin hardware suppliers while Silicon Valley captures all the software engagement and subscription revenue.

By building a truly capable conversational assistant, GM is trying to prove drivers don't need to plug in their iPhones to get an intelligent commute.

However, automotive software carries zero margin for error. A smartphone app can freeze and reboot without putting lives in danger; a car’s core operating system cannot.

There is also the safety question. Studies have consistently shown that holding complex voice conversations while driving still diverts mental capacity, meaning a chattier car might actually increase distracted driving.

Why it matters

GM’s approach here looks remarkably pragmatic.

Instead of overpromising full autonomous driving, it is applying generative AI to immediate automotive pain points: roadside troubleshooting, maintenance diagnostics, and vehicle onboarding. For everyday drivers, an assistant that can accurately diagnose an engine light and book a service slot solves a real, recurring headache.

Strategically, this marks a serious line in the sand against Big Tech. If GM can deliver a voice interface that feels smarter and more native than CarPlay, other automakers will rush to build their own proprietary AI layers rather than handing their dashboards over to Apple and Google.

The real test will be reliability. LLMs are notoriously prone to hallucinations, and inside a two-ton vehicle travelling at 70 miles per hour, giving wrong mechanical advice isn’t just an embarrassment—it’s a safety liability.