Smart home speakers are almost dead. Can AI save them?

• "Smart" devices like the Amazon Echo haven't gotten much smarter in recent years.
• Amazon laid off Alexa workers recently, and Google just announced cuts in its Assistant division.
• But new devices like the AI Pin and the Rabbit R1 may breathe new life into the smart home segment. 

My Alexa has a swearing problem.

I have an Amazon Echo Dot in my kitchen for setting timers and getting the weather. My 7-year-old son also likes to use it for playing music, but this has come with a slew of problems around explicit lyrics.

My son will hear a song on the radio he likes — say, the catchy Bebe Rexha and David Guetta hit "Blue (I'm Good)" — which he has listened to approximately 1,000 times. On the radio, the lyrics go:

"I'm good, yeah, I'm feelin' alright / Baby, I'ma have the best freakin' night of my life."

However, the non-radio edit version uses an expletive instead of "freaking." As you can imagine, this is absolutely thrilling to a 7-year-old boy just learning about the concept of swear words.

But there is no way — no matter how many different iterations and permutations of words I've attempted — to get Alexa to play the clean version of the song on Spotify. Let's just say that more than once, the entire device has gotten unplugged from the wall over my battle over explicit lyrics.

One thing that playing around with ChatGPT and AI tools in the last years has really driven home to me is how stupid Alexa is in comparison; how little it can understand, how rigid and narrow its capabilities are.

Smart speakers seemed so exciting seven years ago, but they've stalled out on innovation — basically, they're just fancy Bluetooth speakers that can also set a kitchen timer. Many people own a smart speaker device and even use it daily, but it hasn't become some magical Jarvis from "Iron Man" assistant or even gotten close.

It's not a stretch to say that the smart speaker category, in its former iteration, is dead.

In 2021, Meta discontinued its Portal video chatting device, the company's foray into the smart home category. At the time, Meta CTO Andrew Bosworth told me that part of the reason was "the entire smart home category has underwhelmed expectations for a while now."

Amazon had layoffs in the departments that handle Alexa in the fall. This week, Google announced more layoffs in the Assistant and hardware teams to allocate more resources to AI.

But suddenly, there's something exciting on the horizon for a smart device you talk to: AI-enabled devices. Something that actually is smart enough not to swear in front of your children.

The AI Pin, a sleek device created by former Apple employees, is a wearable device with a camera. It does things like visually identifying objects and does more complicated tasks, like sifting through emails to tell us about important stuff. It's $699 and aims to replace your phone, which is a lofty ambition, but … hey, it looks cool.

The big hit of this year's CES was the Rabbit R1, a bright red AI handheld device that can take over some of your phone's jobs. You can tell it with your voice to order an Uber or book airline tickets. It has a small screen and buttons, but no camera, and is much more affordable at only $199. (The device sold 10,000 units in its first day of launch.)

Both devices are voice-activated, which is fine at home but more awkward in public, and I'm slightly skeptical socially acceptable it will be to use these at a store or restaurant. I'm also generally skeptical that either of these newfangled devices will catch on widely in these early iterations.

But these new AI devices make me excited about what the future of smart home devices could actually be. A smarter, better version of Alexa — one that actually can do stuff.

I want it to be able to do what I can do inside the Spotify app myself. Stuff that Alexa can't yet do on its own.

I don't just want to have Alexa add toilet paper to my Amazon shopping list — I want to be able to ask it to look up the iCloud location on "Find My" for my husband and see if he's still at the grocery store and then text him that we need toilet paper.

I don't just want a kitchen timer; I want it to read me recipe instructions timed to when each step needs to happen. I want it to be intuitive enough for my younger kid to talk to it.

Most of all, I'm excited harnessing large language models into voice assistants can finally stop my kid from hearing the F-word.