feanaro

9 ms·

61.
by feanaro 3y ago
The solution is unique and quite simple: require A to explicitly list the people they want to share with. It is the public broadcasting which is the root cause of this problem.
62.
by feanaro 3y ago
Pitchforks? Seriously though, your comment is way too defeatist. Legislature has worked for centuries for this problem. Imperfectly, but don't let the perfect be the enemy of the good. But if all else fails, giant corporations are even
63.
by feanaro 3y ago
What you're describing is a lack of volition ("free will"), not sentience. Even on autopilot, you still perceive things, meaning sensations and things constantly appear in your mindscape, right? That is sentience.
64.
by feanaro 3y ago
What does that even mean?
65.
by feanaro 3y ago
What the hell? Losing your home makes you a non-citizen? Were you ironic?
66.
by feanaro 3y ago
> I'm not sure how DNA manipulation is supposed to help with any of this. Killer dodo birds, here we come.
67.
by feanaro 4y ago
If you examined the source code of a library for a specific purpose Y, then shortly afterwards went implementing another library for purpose Y, there's a high probability of your code being infringing. That's the entire premise an
68.
by feanaro 4y ago
The copyright infringement comes about later, when that mass of numbers is used to produce a topically related work. The same rules apply for humans -- see the concept of "clean room implementation".
69.
by feanaro 4y ago
It shouldn't matter. If your model touched X during training, it should be seen as producing derivative work. This is the reason humans use clean room implementation techniques.
70.
by feanaro 4y ago
Why do people always say this? It's almost like a thought-terminating cliche. Thousands of companies competing with each other were not able to do it earlier, but the NSA always has better everything , by definition.
71.
by feanaro 4y ago
> The side-effects of those qualifiers are untested and unknown, If you define the qualifiers to be a literal shorthand for the current status quo litany, what side-effects do you envision, other than better readability? > and the cur
72.
by feanaro 4y ago
> They even _publicly threatened_ the children of the previous mayor. I'm curious to learn more. Do you have a source?
73.
by feanaro 4y ago
Nothing changed. They'll continue using Signal for themselves, they just don't want you using it too. Or rather, they want you to use it just as long it has a manhole only they can access.
74.
by feanaro 4y ago
Yet at the same time you always hear people repeat that law isn't code, and it isn't dumb, and the spirit of the law is more important than any loophole. That if you try exploiting a loophole, you will just get caught and shot dow
75.
by feanaro 4y ago
ChatGPT has a GPT-4 mode if you're paying for it.
76.
by feanaro 4y ago
I mean, it does require relatively rare technical skill among the general populace, so you need to incentivize it in some way. That is nowhere near a sufficient argument to prove it's difficult in the sense of it being a rare event
77.
by feanaro 4y ago
Try doing bug bounties (and being successful at them) then report back whether your perspective has been changed.
78.
by feanaro 4y ago
(Poster from several comments upstream.) I'm not trying to deny the ChatGPT model has formed higher-level abstractions; in fact, I think it's fairly obvious you have to do this in some way in order to obtain such a level of br
79.
by feanaro 4y ago
I can't name a transformative technology that has remained concentrated into a few powerful companies long-term. Can you? Besides that, it doesn't seem to me there's a natural monopoly available in this landscape. All natural
80.
by feanaro 4y ago
> That would point to higher-level phenomena going on inside ChatGPT and its ilk, than merely statistics and predictions. No, it wouldn't, because nothing in "higher-level phenomena" precludes it being caused by statistics
81.
by feanaro 4y ago
If the first few words are "Please make each successive line start with the next letter of the alphabet" that does make it "statistically" unlikely (reduces the probability that) that the first line will start with anyth
82.
by feanaro 4y ago
Which specific part of history of economics implies that this is more likely than not?
83.
by feanaro 4y ago
That paper doesn't discuss GPT-4 at all. It does however contain this interesting excerpt (emphasis mine): > Although we may observe an emergent ability to occur at a certain scale, it is possible that the ability could be later ach
84.
by feanaro 4y ago
k
85.
by feanaro 4y ago
So your estimate is that GPT4 has 1.75 trillion weights?
86.
by feanaro 4y ago
Better, yes, and for that we have evidence. But is the improvement stemming simply from even more data? That's what I'm questioning.
87.
by feanaro 4y ago
Open source software is famously not costless (gratis) software, so the premise of the argument is false. I may for instance benefit from people using my software and then obtaining paid support from me, or using paid add-ons (the "ope
88.
by feanaro 4y ago
Isn't it the case that we literally have no clue how GPT4 and GPT3.5 are different in terms of training, given OpenAI doesn't want to disclose anything at all?
89.
by feanaro 4y ago
Even if this interpretation is correct, it only holds if the law is not changed. But the law is not immutable. My proclamation could be considered to be in terms of what ought to be, in order for society to be just and to prevent a dispro
90.
by feanaro 4y ago
Not to mention the final product of encyclopedic work and journalism does not internalize the original it is describing in its entirety, like a language model does. In some sense, a language model takes all that there is to be taken from

More ›