staunton

51 ms·

751.
by staunton 4y ago
You seem to have very nuanced concepts of "good and bad", which I don't fully grasp. I was starting from a rather simpler world view. Yes, I want society to function. Societal collapse is imho extremely bad. Even regardless o
752.
by staunton 4y ago
Yes. However, I read your post as suggesting (given the quote you respond to) that significant numbers of high school students can/should take university level courses of the same name as their high school courses, presumably because i
753.
by staunton 4y ago
You seem to be saying that schools are bad. Do you really think modern society can function without schools? Or that countries that don't have schools shouldn't build any?
754.
by staunton 4y ago
Would you say that "if a child doesn't want to learn how to read, that's fine, their choice"?
755.
by staunton 4y ago
> There is no way schools can compensate for a bad home life There is a difference between "parents always matter to a child's education" and "let's stop trying to improve schools cause what's the point&quot
756.
by staunton 4y ago
> Linear algebra is not hard Keep in mind that this term covers a huge range of math that can be introduced at very different levels of abstraction and generality. You might think linear algebra is mainly matrix multiplication and Gaussi
757.
by staunton 4y ago
> this is what we will get 10x with free college That seems quite different. First, lots of countries have free higher education and seem to do just fine. Second, lowering price of entry is orthogonal to lowering expected performance. Yo
758.
by staunton 4y ago
Which only proves the point robocat was making
759.
by staunton 4y ago
If you want to succeed in building something useful, there's no way past dealing with risk and accepting solutions that are "good enough". If you have ideas for cool software, this stuff should be the least of your worries. B
760.
by staunton 4y ago
There will never be proof or a 100% effective solution as long as these things are black boxes, which might be "forever". Nor does anyone really need any perfect solutions or proofs. The solution has to be good enough for your pur
761.
by staunton 4y ago
I see, the wording confused me. You don't really "put Fourier transforms around the convolutional layers" because you have to completely replace the convolutions. This seems to be done in some cases. I guess it isn't don
762.
by staunton 4y ago
That's something you want to do if you're trying to build a "personal assistant AI", for example. It has access to most of your data, may talk to others about their dealings with you, and still has to not give away most
763.
by staunton 4y ago
I would go by how well humans do it, which would mean: "yes, you can probably talk it out of it, but when it matters, that's hard enough to do in practice such that the human/system can be used for important tasks"
764.
by staunton 4y ago
> a cursory search suggests it provides the expected speedup What do you mean? Processing a CNN layer takes an amount of time that does not depend on the input data, only the input/output sizes. Fourier transform is just a change of
765.
by staunton 4y ago
I think both of those lose out to the discussion of "grammar in the context of time travel".
766.
by staunton 4y ago
> We are taught to speak and write as concisely and understandably as possible. One day... I believe.
767.
by staunton 4y ago
> The world has evolved such that you need less of it to get by. Not so sure about that. Do you need more intelligence for "working the fields with basic tools" or for an average job in your country? This might change when AI b
768.
by staunton 4y ago
The fundamental error that causes misuse of p-values (and statistics in general) is misunderstanding what statistics is in the first place. Statistics is applied epistemology. There is no algorithm that fits all situations where you're
769.
by staunton 4y ago
There is usually no obvious way to convert the initial condition of an arbitrary quantum system to that for a classical system. The other way around, there is some consensus what the quantization of standard classical systems should be, but
770.
by staunton 4y ago
Just in case you care to know what I'm talking about: I'm pointing out that philosophy had a lot of influence in changing the way science is done (leading to faster progress). This process is ongoing. Of course, you don't hel
771.
by staunton 4y ago
Which is a philosophical criterion. Just 1000 years ago people would have found such a call for "science" strange and deeply insulting to their authority, which they would have needed to dare speculate about things such as math or
772.
by staunton 4y ago
Indeed, LLM's seem to be much worse at introspection than humans. I wonder what would happen if one used reinforcement learning to train into it the ability to correctly predict and reason about it's capabilities and behavior.
773.
by staunton 4y ago
> The tendency GPT and other LLMs to hallucinate is clearly documented. Is this not evidence? "It's not very smart yet" is evidence for "it won't be very smart"? Also, hallucinations are dangerous, at least
774.
by staunton 4y ago
> On a long enough timeline the closed/proprietary approach cannot win. Yes it can. The magic word is "regulation".
775.
by staunton 4y ago
You can mostly convert Python code to Julia line by line (in fact ChatGPT can do it, albeit not very well since there aren't that many Julia examples in the training data). Sometimes you have to look up a function that doesn't hav
776.
by staunton 4y ago
I guess the intended meaning was that you chose your request ("apply what you say to a problem") based on certain philosophical criteria. I think that's actually a surprisingly valid point, given the original (downvoted) comm
777.
by staunton 4y ago
> whether the physical world is perfectly or merely highly consistent What does that even mean? The physical world is not a formal system in any obvious sense. Nor does "consistent" apply to mathematics, by the way, only to for
778.
by staunton 4y ago
Yes, good notation helps dealing with complexity. My point was that "number of equations" involved in a model or theory is not an objective measure of, really, anything. Every system of equations can be written as "A = 0&quot
779.
by staunton 4y ago
Number of equations != How simple something is
780.
by staunton 4y ago
Computational complexity is defined for programs with respect to a parameter in the limit where that parameter is large. What's the objectively correct parameter for this comparison that every theory of physics has, after you convert i

More ›