GM is shoving Gemini into four million cars

GM just announced they’re pushing Google’s Gemini AI assistant into roughly four million vehicles across the US. That covers model year 2022 and newer Cadillac, Chevrolet, Buick, and GMC cars that already have Google built-in. The rollout happens over several months via over-the-air updates to the infotainment system.

GM is calling this “one of the largest deployments of Gemini in the industry.” That’s probably true, but it’s also the kind of stat that sounds impressive until you realize it’s just a software toggle for cars that already had Google Assistant. The real question is whether Gemini actually makes things better or just adds more complexity to an already bloated dashboard experience.

According to GM, customers will “notice an upgrade from the current Google Assistant to a smarter, more intuitive AI assistant that continues to improve over time.” I’ve been using Gemini on my phone, and while it’s definitely smarter than the old Assistant in some ways—better at multi-step queries, more natural conversation flow—it also hallucinates more and sometimes takes longer to respond. In a car, latency matters. You don’t want your voice assistant pausing for three seconds while you’re trying to navigate a merge.

I’m also skeptical about the “continues to improve over time” line. That usually means more data collection and more updates that could break things. GM’s infotainment systems have historically been laggy and prone to crashes. Adding a heavier AI layer on top feels like asking for trouble.

Still, I’ll give them credit for scale. Four million cars is a lot of real-world testing. If Gemini handles navigation, climate control, and media playback reliably, it could actually be useful. But if it turns into another half-baked voice feature that mishears commands or struggles with background noise, it’ll just be another thing drivers ignore.

I’ll reserve judgment until I can try it in a rental. For now, this feels like GM checking a box on a buzzword bingo card rather than solving a real problem.