This seems like a ton of hype for a marginal (if that) improvement over existing art.
We've had "truly random" numbers for a long time, starting with those based on radioactive decay.
Current hardware RNGs already take advantage of provably random physical processes. Intel's RDRAND implementation is one good example. Yes, it is probably not a "perfect" source of randomness, though one can always take other sources of entropy and mix further. Doing so will always move the output towards a more random state.
I find it unlikely in the extreme that some future tech will be able to "predict" values from such a random stream (like you could with a PRNG given the seed). In other words, your cyphertext based on a RDRAND+ generated one time pad is safe for eternity, unless the pad is compromised.
Most commenters here are focused on the Microsoft hardware, but the article mentions systems from several other manufacturers using the same hardware. Some of them, Dell and Lenovo in particular, have good Linux support.
Nvidia has also supported Linux well in general, so let's hope there's an attractive Linux option soon!
Personally I'd be just as happy with a small form factor desktop with the same hardware.
I'll have to take a look at OxCaml. I'm leery of "C-like" performance claims after Java has thoroughly failed to live up to a similar claim after thirty plus years of development... What it's actually achieved is about 50% C performance, IF you're willing to give it a huge heap, at least 2x the actually required memory.
Rust is clearly well positioned for deeply embedded work, and has actual C/C++ level performance. Given AI coding assistance, Rust is looking more and more approachable...and of course faster processors and compiler improvements will solve the compilation speed issue over time.
All that said, there's nothing wrong with a fast, safe language with ML syntax!
(One dark horse in all this is Mojo, which may provide Rust level safety with a more ergonomic language, and a much faster compiler...)
The disincentives to nuclear war are glaringly obvious enough that even politicians (and their masters) get it.
AI isn't like that. One problem is that it's rather generally misunderstood at this point. "AI" is not "intelligence". It's intelligence-adjacent, and something like LLMs is part of our psyche...the subconscious facility that allows us to form sentences without really thinking about it.
At any rate, I have to agree with most of the points the blog author brings up.
I agree that if the best we can do is something that can't be self-sustaining, Mars should wait until that changes.
I disagree with KSR's main points. Perchlorates are solvable, the effects of Martian gravity are not known (and are solvable if there is a problem), and finally radiation is a non-issue for those living in the only sane place on Mars, underground.
Whether or not Mars is a target in the near term, we need to proceed with our current plan of establishing a permanent base on the Moon. The only way to improve on Earth's resource limitations is to exploit the virtually unlimited riches available beyond her atmosphere, and the Moon is the first step. It's also a great place for heavy industry, not to mention astronomy!
LLMs are amazing technology. It's crazy to interact with something that knows a lot about effectively everything that's ever been written, as well as mimicking human cognition to a large degree.
What LLMs are NOT is intelligent in the same way as a human, which is to say they are not "AGI". They may be loosely AGI-equivalent for certain tasks, software development being the poster child. LLMs have no equivalent of "judgement", and they lie ("hallucinate") with impunity if they don't know the answer. Even with coding, they'll often do the wrong thing, such as writing tests that don't test anything.
It seems likely that LLMs will be one component of a truly conscious AI (AGI+), in the same way our subconscious facility to form sentences is part of our intelligence. We'll see how quickly the other pieces arrive, if ever.
According to current theory AIUI, cosmic inflation greatly influenced the CMB. It ended approximately 10^-32 seconds after the Big Bang:
"Cosmic inflation is believed to have occurred in an incredibly brief, rapid, and exponential expansion phase lasting from approximately 10^-37 to 10^-32 seconds after the Big Bang. During this minute interval, the universe expanded by a factor of at least 10^26, and potentially as much as 10^50."
I'm surprised the author of this article thinks Go is a "system language".
Go uses GC, and therefore can't be used for hard real time applications. That's disqualifying as I understand it.
C, C++, Rust, Ada, and Mojo are true system languages IMO. It is true that as long as you can pre-allocate your data structures, and disable GC at runtime, that GC-enabled languages can be used. However, many of them rely on GC in their standard libraries.
I think there are some major problems with this thinking. How does this relate to human artists who studied prior art and then produced something?
I’ll grant you that AI isn’t actually intelligent, but I’ve seen many images and video that exhibited a good bit of originality, and were at a minimum a derived work…
One reason foreign enrollment is declining is concern about (mainly) Chinese espionage. That’s entirely reasonable, given the vast amount of stolen engineering and research…
We've had "truly random" numbers for a long time, starting with those based on radioactive decay.
Current hardware RNGs already take advantage of provably random physical processes. Intel's RDRAND implementation is one good example. Yes, it is probably not a "perfect" source of randomness, though one can always take other sources of entropy and mix further. Doing so will always move the output towards a more random state.
I find it unlikely in the extreme that some future tech will be able to "predict" values from such a random stream (like you could with a PRNG given the seed). In other words, your cyphertext based on a RDRAND+ generated one time pad is safe for eternity, unless the pad is compromised.