I suspect that chances of would-be burglars or identity thieves breaking into Google data are pretty slim, in comparison to a home-installed system.
OTOH both Google and a private person can be strong-armed by a court order, or even a three-letter agency, to open up their AI knowledge vaults.
It's like saying "People would never break into a bank, when they could break into someone's house and steal their stuff"
It may be easier to break into someone's home-brew system, but generally it would be unlikely to happen unless you were being otherwise targetted. Whereas google has a lot of users data, which could make it a more attractive target.
Just how "home-brew" are we talking here? If there's any web-facing code that you didn't write yourself, whether commercial or open-source or whatever, that's a target for attackers that just scan everything looking for known vulnerable services.
If it is entirely custom, I'm fairly sure there are a few classes of common security errors that can be reasonably well tested for without direct human involvement. Which brings back the threat of attackers just scanning for all available targets.
If you're a "person of interest", you'll be attacked either way.
If you're just a regular person, like me, you won't likely be targeted. But my chances of getting a collateral damage are much higher in a centralized system: when it's broken into, my data have a chance of being siphoned off along with data of someone less ordinary, who was the target of the attack.
But there's a third case: "normal boring people" become interesting just by being together in a big group, even if no particular "person of interest" is among them.
The odds of someone physically taking my computer and decrypting my data are zero (128-bit encryption for the win!).
The odds of someone electronically breaking into my computer are higher.
The odds of my data being misused or misaccessed at Google are one: it will happen.
Maybe someday that will be a realistic endeavor, but it would take a lot of effort to set up and maintain your own personal versions of all of google's services, and integrate them
But it stores the private data about your location, your searches, your purchases. this could even be an encrypted private fire-walled bit of cloud rather than a robot in your house.
The point being your AI is serving you. And you can delete/edit this private data (or even the whole AI) if you wish.
https://en.wikipedia.org/wiki/Intelligent_agent
Perhaps using Urbit as the network architecture since it already has pki. As Yarvin calls it, true 'personal cloud computation'.
Really it's the only sane way to live in the 21st century.
I've designed my home automation system on the concept that the only route to the Internet is my computer (and therefore, my home automation software). My computer is a secure, well-managed intermediary that can store my data, and decide when and how to receive and send data to the Internet.
The idea of dozens of Internet-connected devices in the home is :terrifying: in comparison, especially considering that badly-secured IoT devices are now powering some of the biggest botnets out there.
My light switch, however, cannot talk to the Internet. It has local-only communication protocols that are simple. It knows how to be told to turn the lights on or off or dim or a handful of other settings, but it's literally incapable of doing anything else... and why would I want anything different? Why should my light switch have Bluetooth and Wi-Fi and software updates and a miniature flavor of Linux... It's a switch!
When my AI can talk directly to your AI, we can transfer mail etc without cloud services. If it knows our social network, perhaps it can remotely store encrypted backups of our data, but only with people who we already trust.
It's just not easy to do, both in setup and maintenance effort. Most people would not care about their privacy nearly enough to justify that.