staunton

17 ms·

211.
by staunton 2y ago
They just have no space left in the jails, what can you do... I guess they hope that as long as protesters get a spot the damage to society will be manageable.
212.
by staunton 2y ago
> Try publishing a paper on fine-tuning LLMs with the goal of creating agents capable of replacing the executive suite and the board of directors for major corporations and see what happens to your career. I expect if you manage to get s
213.
by staunton 2y ago
> Is this article roughly the application of the startup mindset to research? Somewhat but a few crucial points are missing for it to be that. For example, - hire a lot more people (PhD candidates, postdocs) than you have funding to pay
214.
by staunton 2y ago
I don't understand what the scam is. The structure is opaque which indicates there might be some scam but how would one know what it is?
215.
by staunton 2y ago
> leaving some things like performance and size limits deliberately unspecified in API's seems like more of a feature than a bug In rare cases there might be exceptions. Hard real time applications and constant time cryptography com
216.
by staunton 2y ago
> Zig is, right now, being used for high-assurance systems I'm not convinced this is telling us very much. I was talking about software that, e.g., causes deaths when it fails. But regardless of what level of assurance one looks at,
217.
by staunton 2y ago
It's perfectly feasible to have proofs about programs together with a guaranteed upper bound on (something like) the "number of processor instructions it will take" (given known finite bounds for all inputs). Of course, just
218.
by staunton 2y ago
> Rust is starting to show some warts (despite still being a massive improvement over C from a safety perspective) The way it seems to me is that actually Rust aims to be an improvement over C++ rather than C. (And Zig aims to be an impr
219.
by staunton 2y ago
> Work on porting mathlib to Lean 4 is far along As far as I understand, that work is in fact done .
220.
by staunton 2y ago
> a drone that is capable of detecting targets reliably on its own Yes, I guess in the future a lot pf people will be killed by autonomous swarms of drones that just don't like the way they look, sometimes malfunction, give plausibl
221.
by staunton 2y ago
> we will only see their results when smart missiles start raining on our heads. I'll remember to say that whenever anyone wants to apply any law or regulation to anything I'm doing
222.
by staunton 2y ago
Never let "good" be the enemy of "cheaper and roughly on par with the competition".
223.
by staunton 2y ago
You're the second person who understood the opposite of what I said. I'm not a native so can you tell why what I said was ambiguous or contrary to my intended meaning?
224.
by staunton 2y ago
> Think how much value is locked up behind cryptography. It's actually "so much value" that it's not possible to take much of it if you reveal that you did. People can "fork (parts of) the financial system"
225.
by staunton 2y ago
> the safest path forward is to assume encryption is broken and move to post-quantum crypto before we “need” to. An even safer path would be to use both "classical" and post-quantum methods in combination ("hybrid").
226.
by staunton 2y ago
That's exactly what I said?!
227.
by staunton 2y ago
> The method and actor are irrelevant. That may be what you mostly care about, but it is definitely not true in deciding legal questions.
228.
by staunton 2y ago
For example, look at the fines German car manufacturers "faced" in the EU when their emission-cheating scheme was exposed...
229.
by staunton 2y ago
It's much harder to fine companies that break the law if they make up a substantial part of your economy. On the other hand, big US companies don't have as much lobby power in the EU, so the EU is "free" to fine them.
230.
by staunton 2y ago
What would you say to someone who claims that "never" obviously means "not before the end of this quarter, cause who could possibly care about anything beyond that"?
231.
by staunton 2y ago
> We've had advanced LLMs for years now and nobody has summoned Yaldabaoth Admitting that as an argument, there is basically nothing that should ever worry you. For example: Climate change? We've been blowing stuff into the a
232.
by staunton 2y ago
Real example: when you're working on the data analysis of one of the "beyond the standard model" physics experiments. For example, there is one where they basically shoot a big laser against a wall and see if anything goes th
233.
by staunton 2y ago
> You can prove it doesn't crash, it doesn't leak memory, it doesn't break authorization invariants, etc. You can but nobody does. It's enough if it doesn't usually do these things.
234.
by staunton 2y ago
> Not sure why this kind of "we are aiming for AGI" code is written in Python Python is used because it's easy to use for these high-level numeric calculations (compared, e.g., to C) and all the relevant libraries are avai
235.
by staunton 2y ago
A hypothetical good AI would have reviewed the experimental design and pointed out the misconception
236.
by staunton 2y ago
If we had scalable criteria for quality of papers, that would be the end of science. The rest would be engineering...
237.
by staunton 2y ago
What's a "standard flag"? Couldn't find anything about it... I would think "upper triangular" would be a weird/bad notion to talk about in a basis-independent setting (because it depends on the basis).
238.
by staunton 2y ago
The NSA had the option to do something like that when they (via NIST) standardized DES. They chose to standardize a version that's secure against attacks that only they knew at the time, shorten the key length so they can still brute-f
239.
by staunton 2y ago
It's to do with "if we ever have big/good quantum computers, prime factorization is doable" and with "let's use the new shiny things". On a related note, someone might discover some elliptic curve math tom
240.
by staunton 2y ago
If you mistrust authors and can't actually rule out someone trying to cheat, checking a formaized proof beyond a reasonable doubt is still a lot of work and needs a huge amount of domain-specific knowledge. An author trying to cheat wo

More ›