331–340 of 408 posts
Earlier quoted context omitted.
> It's trivial to add a matrix to account for neutrino masses The matrix you are thinking of is presumably the PMNS matrix [1]. It's equivalent to the CKM matrix for quarks [2]. The purpose of both is to parametrize the mismatch between flavor [3] and mass eigenstates, not "to account for neutrino masses" or "explain their origin". As far as the standard model is concerned, neutrino masses and quark masses all origin…
Thanks Lord Kelvin
Earlier quoted context omitted.
The additional irony here is that LLMs are a tool that is likely forever damned to regurgitate knowledge of the past, with the inability to derive new information.
It depends on what you mean, specifically on your distance metric. If you mean nearest neighbours search like autocorrect then LLMs are extrapolative. You can easily generate combinations not seen before. I mean you can prove this with parametric prompting. Like "Generate a poem about {noun} in {place} in {language}" or whatever. This is a simplistic example but it doesn't take much to come up with a space that has q…
But I get it, the interpolation you’re talking about is limited. But I think you missed this insight: human interpolation is limited too. In the short term everything we do is simply recombination of ideas as you put it.
But that’s the short term. In the long term we do things that are much greater. But I think this is just an aggregation of small changes. Change the words in a poem 5000 times: have the LLM do the same task 5000 times. Let it pick a random word. The result is wholly original. And I think in the end this what human cognition is as well.
Earlier quoted context omitted.
Your conception of an “explanation of reality” is deeply flawed.
you can correctly predict reality whilst having absolutely know idea how it works (ie the path of a photon in the double slit experiment).
The “What path did the photon take?” question is one of those times. The answer to the question is Mu.
Similar to the questions “How much phlogiston is there in iron?” or “Does sulphur have more earth than air, or more air than earth?”.
Experimental particle physicist here. It's just hard. I measured the electron's vector coupling to the Z boson at SLAC in the late 1990s, and the answer from that measurement is: we don't know yet - and that's the point. Thirty years later, the discrepancy between my experiment and LEP's hasn't been resolved. It might be nothing. It might be the first whisper of dark matter or a new force. And the only way to find ou…
>"It might be the first whisper of dark matter" Come now.
Many of those models naturally include a dark matter candidate. I didn't mean to imply 'we found dark matter' — it's that the theories which could explain the discrepancy often come with one attached.
Earlier quoted context omitted.
Better superconductors here. Would you like a $20 MRI down at your local drug store to detect cancer at early stage 1?
The problem isn't the cheaper MRI. The problem is the expert that needs to interpret the results. Detecting millions of cancers that don't actually exist doesn't help anybody.
Earlier quoted context omitted.
Why can't some of these trillion dollar companies invest back in the quantum tech that got them there, if it's so certain there will be benefits? Why not Apple and Nvidia fund the next particle collider, and give something back to society instead of letting tax payers fund it so billionaires can privatize the profits?
Fundamental physics research has an extremely profitable returns ratio, but it takes decades to amortize. This does not work with capitalist corporations who only care about immediate profits. Even for governments this is a difficult sell, but at least they don't have to soothe shareholders every quarter. Generational projects take a different kind of economic thinking.
If that's the case it seems like it might be shrewd for younger investors to buy into physics research on a 15-20 year timeline?
Earlier quoted context omitted.
> This kind of slow, incremental improvement that costs tens of billions of dollars and takes decades gave us the microchips that ultimately enabled you to type this comment on your phone/computer. No. These two cases are absurdly different, and you're even completely misunderstanding (or misrepresenting) the meaning of the "tens of billions of dollars" figure. Microchips were an incremental improvement where the ind…
> Absolutely not Engineers not being able to fathom that by building this huge-ass and complicated machines to answer questions about the fundamentals of nature, other problems are solved or new things are invented that improve and change our life will never not be funny to me
Earlier quoted context omitted.
Doesn't that imply our theories are "good enough" for all practical purposes? If they're impossible to empirically disprove?
If I have to make a guess, we are at the level of pre-copernicus in particle physics. We are finding local maximums(induction) but the establishment cannot handle deduction. Everything is an overly complex bandaid. At some point someone will find something elegant that can predict 70% as good, and at some point we will realize: 'Oh that's great, the sun is actually at the center of the solar system, Copernicious was…
This sounds like a distinction without consequence, but I think that's wrong. The sun is not special. It just has a lot of mass. If somebody learns: The earth orbits the sun-- They don't understand how two black holes can orbit each other. If somebody learns: The sun and the earth orbit their CM -- They will be able to understand that.
Earlier quoted context omitted.
quantum mechanics (also very much not Newtonian) is much more important to our day-to-day lives.
this kind of distinction is quite stupid in general as plenty of things that we rely on for day-to-day activities such as our houses, desks, chairs, beds, shoes, clothes, etc are all based on Newtonian/classical mechanics. Basically everything that we use which existed pre-transistor strictly speaking only required classical physics.
It's basically the opposite situation from 150 years ago. Back then, we thought our theory was more or less complete while having experimental data which disproved it (Michelson-Morley experiment, Mercury perihelion, I am sure there are others). Right now, we know our theories are incomplete (since GR and QFT are incompatible) while having no experimental data which contradicts them.
>GR and QFT are incompatible I did physics at uni and kind of dropped out when it got too hard. I've long guessed the incompatibility is because the maths is just too hard for human brains, though I'm probably biased there, and we'll get a breakthrough when AI can handle much more complex maths than us. Probably not so long till we find out on that one. I once tried to write a simplified explanation for why a spin-2…
I think that even if AI were to find a good unification of GR and QM, we wouldn't be able to test it. We might accept it without additional confirmation if it were sufficiently natural-feeling (the way we accepted Newtonian gravity long before we could measure G), but there's no guarantee that we'd ever be able to meaningfully test it.
We could get lucky -- such a theory might point at a solution to some of the few loose threads we get out of existing collider and cosmological measurements -- but we might not. We could be stuck wishing we had a galaxy-sized collider.