I picked one at random (B.2 -- the second autocorrelation inequality). Then, I looked up the paper that produced the previous state of the art (https://arxiv.org/pdf/0907.1379). It turns out that the authors had themselves found the upper bound by performing a numerical search using "Mathematica 6" (p.4). Not only did the authors consider this as a secondary contribution (p.2), but they also argued that finding something better was very doable, but not worth the pain:
"We remark that all this could be done rigorously, but one needs to control
the error arising from the discretization, and the sheer documentation
of it is simply not worth the effort, in view of the minimal gain." (p.5)
So at least in this case it looks like the advancement produced by AlphaEvolve was quite incremental (still cool!).
Is there a good scientific/safety reason for which Tesla's batteries are held together by polyurethane cement, as opposed to something easier to deal with?
NPEs help make the market for patents more efficient -- small firms invent and have a hard time protecting their patent? They sell it to an NPE.
Is the system imperfect? Probably. Worst case scenario, a big firm get frivolously sued and loses -- this is just a transfer of money from Apple to the NPE and the small inventor. Will Apple be less likely to invent as a consequence? I don't think so, but you can prove me wrong.
Best case scenario, small actors are incentivized to invent and sell their ideas in the market for patents.
So why is everybody outraged? Because the lawful licensing agreements don't make it to the news. The outrageous cases do.
Draw three independent random variables a,b,c~U[0,1].
In the first experiment, let the baby pick between a and b: of course E[a|chosen]>1/2; E[b|unchosen]<1/2. Since a,b,c are independent, E[c]=1/2.
Why are they (journalist/researcher (?)) surprised that in the 2nd round the baby chooses c over b?
[Of course, if you force the baby to pick either b or a in the first round, she will be equally likely to pick c or not in the second round]
Does someone have a good explanation/intuition for why you cannot exploit quantum entanglement to send information faster than light?
If me observing the particle in Australia alters the probability distribution of your particle in USA, can't I only observe the particle when I want to communicate 1 and never observe it when I want to communicate 0?
Edit: thanks a lot for the answers! I guess it boils down to the fact that the Australian guy cannot condition his decision on the (unknown) spin of his particle -- if he could (eg: had access to the local hidden information) then he would be able to update the USA's probability distribution instantaneously and use it to communicate
People are commenting that the FB's bureaucratic system is idiotic and unsustainable.
I don't understand why: bureaucracies are a brilliant response by oversized institutions/businesses to force people to reveal how much they care about solving an issue.
Don't really need your website to be cleared? You won't make a fuss about it.
Is it unjust and inequitable? Yes. Is it unsustainable? Cannot see why.
No bad assumption here: people say that productivity growth and wage growth ought be the same. If this was not the case, then in the long run 100% of value added would go to capital!
Of course this does not mean that the productivity gains should directly translate into wages.
CEO wage depends on company size: big companies are willing to pay A LOT for a CEO which screws up with 1% less chance.
Other jobs (be it janitor, software engineer) don't scale up as much with company size, since the single tasks remain the same.
So a big reason CEO wages grew is that companies became massive.
Great comment: more in general, dual licensing favors (or at least, does not stand against) the development of proprietary software -- so it may be ok for current users of the GPL licensed product, but it definitely impacts the direction of innovation
Agreed. I'd be inclined to blame the coverage rather than the researchers for the 'coarse' interpretation (after all, the original title talks about a 'neural correlate' of consciousness)...
I do not understand why this is about consciousness. My take of this study is that they establish that there are two type of neurons: (i) those recording whether there is a signal ("neurons signalling stimulus intensity" (ii) those recording how to react on the signal based on a rule ("representing the crows' percept").
This is cool, but what does this to have to do with consciousness?
They mention that they're not sure either about "phenomenal consciousness" and "access consciousness", but I wish they elaborated further on this.
But how incremental are these advancements?
I picked one at random (B.2 -- the second autocorrelation inequality). Then, I looked up the paper that produced the previous state of the art (https://arxiv.org/pdf/0907.1379). It turns out that the authors had themselves found the upper bound by performing a numerical search using "Mathematica 6" (p.4). Not only did the authors consider this as a secondary contribution (p.2), but they also argued that finding something better was very doable, but not worth the pain:
"We remark that all this could be done rigorously, but one needs to control the error arising from the discretization, and the sheer documentation of it is simply not worth the effort, in view of the minimal gain." (p.5)
So at least in this case it looks like the advancement produced by AlphaEvolve was quite incremental (still cool!).