3 ms·

I see a lot of comments here about how private AI is much cheaper than doing this, and that is currently true. But I am curious if anyone can think of any use

I see a lot of comments here about how private AI is much cheaper than doing this, and that is currently true. But I am curious if anyone can think of any use cases for this as it currently stands? For every idea I can think of, I can think of another solution that would probably be a better solution at current slowdown in cost from using homomorphic encryption. For their two examples... 1) For the "will they click the add" one, how would anyone verify that the model I am uploading of myself is remotely accurate? And can't they learn my characteristics by running them against very specific ads that pinpoint my preferences? 2) For the "Do you have heart disease" one, why not just generating a billion datapoints (or more). The user can download the compressed data of this huge sample of points and then just enter their data on their own side and find the point that is closest (if not an exact match) to their own data? I'm not pulling apart the idea of homorphic encryption, I'm just curious if anyone can think of a really compelling use case for it given it 10^3 to 10^6 greater compute cost?

catlifeonmars 1mo ago
If the bottleneck is data transfer rate or volume, then I think you’ll find more use from today’s FHE. Incidentally there is a fair amount of overlap in the use case space between homomorphisms and oblivious transfer.
- stephen_cagle 1mo ago
  Huh, that is an interesting thought. My belief is that a FHE scales at least linearly to the size of the memory. My belief for this is that it if didn't I could "learn" what computation you are doing by analyzing what parts of memory you did not look in. So, by that logic, we are potentially looking for a problem space where their is a large amount of data that is too large to even be paged through by the client, and where the FHE would actually make financial sense relative to downloading it? I'll think about that, but nothing immediately comes to mind.

catlifeonmars 1mo ago

If the bottleneck is data transfer rate or volume, then I think you’ll find more use from today’s FHE. Incidentally there is a fair amount of overlap in the use case space between homomorphisms and oblivious transfer.

stephen_cagle 1mo ago
Huh, that is an interesting thought. My belief is that a FHE scales at least linearly to the size of the memory. My belief for this is that it if didn't I could "learn" what computation you are doing by analyzing what parts of memory you did not look in. So, by that logic, we are potentially looking for a problem space where their is a large amount of data that is too large to even be paged through by the client, and where the FHE would actually make financial sense relative to downloading it? I'll think about that, but nothing immediately comes to mind.