Let me ask you a serious question: How would you limit the construction of AI data centers VS regular ones? Or even a data center inside of an existing office building? Or a data center that's just a backup location (where half of the equipment might not even be turned on all the time)?
How do you enforce such a thing? I could build a totally normal data center, then install a bunch of Nvidia HBMs and no one would know. You'd need a specific legal entity, the Data Center Police that dictate how, when, and why any given type of hardware gets used.
On the face of it, it's a huge violation of freedom of speech. Because you'd have to look into what specific software and data was being run/used on the hardware to see if it's related to AI and not, say, a CGI rendering farm. Or protein folding. Or physics/medical research. All of which use the same exact hardware.
I'm not saying you can't regulate data centers. You just can't regulate what people do with any given hardware.
My recommendation: Regulate data centers in these ways:
- They must be powered by local renewable energy. None of this, "we bought carbon credits" bullshit.
- If they need water cooling, they must use water from local retention ponds. Not the local potable water supply (using reclaimed water would be OK though).
Implement those two requirements and no one would have any basis to be bitching about data centers, specifically. There's much, much worse business/industrial buildings than data centers.
Aside: Data centers are just climate-controlled office buildings (with extra cooling). They just sit there. From a noise perspective, they're on par with a warehouse. In fact, warehouses that stay open 24/7 are much, much worse because of the truck traffic and forklifts. While a data center is a 24/7, low-frequency hum (on the outside, nearby), a warehouse is going to be shaking things nearby.
...and nothing is worse than busy train tracks! I slept at a friend's place for a few days that lived next to an elevated train (across the street from his window). There's no way a data center is worse than that! No way.
The doc would have no clue what his local IT staff do with his data, either. Furthermore, for all you know, the doctor isn't deleting the things they record. In all of these scenarios, it's 100% out of your control. Even if it's local. Because it's not your hardware and it's not your business (as in, you don't own the doctor's office/hospital).
What you're complaining about is contract law. You obviously don't trust it. That has nothing to do with your voice being sent to servers in an Amazon data center. The 3rd party transcription service provider is under contract not to share your data and so is Amazon. Just like the doctor's office is required by law not to share your data without your consent (which I guarantee you gave them permission for when you did your paperwork at that office).
Now let me take a step back for a second and ask a more serious question: Why do you care? I mean, I can think of several reasons why someone could care about theirs and their doctor's voices being recorded but why do you, specifically care? What's the "this is a problem" scenario you're worried about?