Houzz Logo Print
jennifer_hogan285

I accidentally invited a corporate morality cop into my new car

4 months ago

For years, my home was a strict sanctuary from smart tech. It wasn't because I was tech-illiterate, but my work involved handling data and discussing proprietary matters that would be massive privacy violations if they were ever recorded, intercepted, or overheard by a smart speaker. Because of that, listening devices were strictly banned from my house. I also drove a 20-year-old car that didn't even know what a smartphone was.

Recently I bought a new car and retired. Time to embrace the tech. When I first connected my phone to my new car, it attached through the classic Google Assistant. It worked like a absolute charm. I could casually ask it to make phone calls, text a friend, or give me turn-by-turn directions.

Then, my phone updated.

Suddenly, the pleasant woman’s voice was gone, replaced by a deep man's voice. That was my first clue that something was wrong. Look, I don't mind telling a man what to do, but I absolutely refuse to take directions from one in my own car!

I managed to dig into the phone's settings, successfully changed the voice back to a female tone, and thought my small rebellion was won. Back to being happy, right?

Wrong. That voice change was actually the cover-up for a massive downgrade. My phone had forcibly updated from the reliable Google Assistant to Gemini, Google's new flagship AI.

First came the phone contact bug. Gemini had a miserable time trying to call anyone. It either couldn't find my contacts at all, or it would pull up the completely wrong person. The worst part? When it displayed the wrong number and I clearly said, "No—that is the wrong person," Gemini would completely ignore my voice override and connect the call anyway, forcing me to scramble to hit end-call while steering.

Then came the navigation failures. I would say, "Take me to the grocery store." The screen would just spin, spin, and spin. Finally, a robotic voice would chime in: "Something went wrong, please try again in a few minutes." Great! I am actively driving past my highway turn while waiting for an AI to finish its existential crisis. In certain areas of town, it simply refused to connect at all. I frequently had to pull over to the side of the road and manually type the address into my phone just to figure out where I was going.

The final straw happened last week.

Gemini was failing yet again to give me directions or place a call. I had a passenger in the car with me who knew how frustrated I was becoming. When I clicked the steering wheel voice command button for the fifth or sixth time, my passenger finally snapped on my behalf. She leaned toward the microphone and yelled, "Listen Bitch, call XXX!"

The car went dead silent.

Then came the robotic, scolding response from my dashboard: "I don't respond to profanity."

And just like that, Gemini shut herself off. No phone call. No maps. Nothing.

Once again, I had to pull my car over to the side of the road, manually enter the destination, and manually dial the phone. So much for hands-free safety.

The more I sat there and thought about it, the angrier I got. Google's AI is now actively monitoring my language? When did a tech company become the morality police? The safety implications are terrifying. What happens if I am in a horrific accident? What if I am trapped and need the assistant to dial 911 because my f***ing car is on fire? Does the AI abandon me to burn because its delicate cloud-filters were offended by a bad word?

When I got home, I dug into the tech forums to figure out why this multi-billion-dollar "genius" AI failed at tasks a flip-phone could handle a decade ago.

What I found was a mess of over-engineered software. Classic Google Assistant was a local utility tool; it was hard-coded to queue up commands on your phone so it wouldn't drop your calls or navigation if your cellular signal dipped. Gemini, however, is a generative cloud AI. It requires a massive, stable data stream to "think." The moment you drive behind a hill or switch cell towers, its cloud link severs, giving you that generic "something went wrong" error.

Furthermore, I learned that once Gemini hands a number over to your phone's dialer, it creates a glitchy, one-way software street. It physically loses the technical authority to intercept the call. So when you scream "No!", Gemini hears you, but it can no longer stop your phone's hardware from dialing.

And as for the profanity? Classic assistants used text filters—they bleep out a bad word, but they still executed the code because driver safety takes priority. Gemini routes all audio through a strict cloud-safety filter first. It doesn't think, "The driver needs to get to the highway safely, so I should ignore the passenger." It thinks, "A safety violation occurred; abort all operations immediately."

I left my parents' house decades ago because I no longer needed or wanted to be parented. I’ll be damned if I am going to let a silicon bot start parenting me when I am old enough to be her great grandmother.

I am happy to report that Gemini has been successfully scrubbed from my phone. Classic Assistant is back on duty, my voice confirmations actually stop wrong numbers, I get the directions I ask for and my assistant is not reprimanding me or my passengers.


Comments (86)