yeah, for a moment I was reading it as being a holomorphic encryption type setup, which I think is the only case where you can say 'provably private'.
It's better than nothing, I guess...
But if you placed the server at the NSA, and said "there is something on here that you really want, it's currently powered on and connected to the network, and the user is accessing it via ssh", it seems relatively straightforward for them to intercept and access.
would you be willing to guarantee that some automation process will never mess up, and if/when it does, compensate the user with cash.
For a compiler, with a given set of test suites, the answer is generally yes, and you could probably find someone willing to insure you for a significant amount of money, that a compilation bug will not screw up in a such a large way that it will affect your business.
For a LLM, I have a believing that anyone will be willing to provide that same level of insurance.
If a LLM company said "hey use our product, it works 100% of the time, and if it does fuck up, we will pay up to a million dollars in losses" I bet a lot of people would be willing to use it. I do not believe any sane company will make that guarantee at this point, outside of extremely narrow cases with lots of guardrails.
That's why a lot of ai tools are consumer/dev tools, because if they fuck up, (which they will) the losses are minimal.
I think there is a degree of trade off in this, yes a nuclear scientist/engineer/technician in Russia or China is cheaper than in the USA. But also, the people with those kind of skills (or those technically competent enough to do a good job, are going to be expensive no matter what.)
At some level when people have enough technical skill to do these jobs well, they also have enough technical skill to leave the country and go elsewhere and do something else for better quality of life.
Like GDP per capita in china is much lower than the USA, I bet that their nuclear program engineers are getting paid at least ~80k range, which while less than the equivalent engineer in the USA is paid, is not the same level as what a direct PPP comparison would give.
One of the best benefits of the current no live nuclear testing treaties / environment, was that the United States was one of a few countries that had done extensive live tests early on.
The United States is able to sit on its arsenal and data, and with extensive research and simulation validate to a high degree of accuracy that "hey our bombs still work".
Most countries do not have the data/technical expertise/resources to be able to validate with just simulation. But since no-one else is doing live tests, they do not do live tests either.
How much do you want to bet that a subset of the Russian nuclear weapons simply do not work, and that they will only figure this out when they need to 'test' in response to American tests.
It's better than nothing, I guess...
But if you placed the server at the NSA, and said "there is something on here that you really want, it's currently powered on and connected to the network, and the user is accessing it via ssh", it seems relatively straightforward for them to intercept and access.