Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Electronic Structure of LK-99

arxiv.org

Post reply on HN
Print view View original

411–420 of 448 posts

Re: Electronic Structure of LK-99

#411

Post by TechBro8615 »

appplication wrote: post #31
vesinisa wrote: post #26

Earlier quoted context omitted.

Just curious, how did you find Covid "exciting"? Making lots of money on the bear market?

Perhaps poorly worded, I meant since before . During Covid I lost a lot of zest for life, felt like I went to sleep in my 20s and woke up an old man.

I felt the same, but I also feel like it was the perfect (least bad) time in my life for it to happen. I can't imagine being in the middle of High School or University during Covid - those kids got the absolute worst end of the deal. I was also recently single, and counted my days feeling grateful I was not living with my ex when the pandemic hit.

Re: Electronic Structure of LK-99

#412

Post by jacquesm »

ifyoubuildit wrote: post #410
jacquesm wrote: post #400

Earlier quoted context omitted.

Science served us well enough that we're all still commenting here. It's the prevention paradox all over again, if science had not done its work you'd be in a much better position to appreciate the difference that it - and healthcare - made but then you might not have made it at all... science definitely did not let us down. Who did let us down: the people that were making things worse from day #1 by stirring the pot…

> Who did let us down: the people that were making things worse from day #1 by stirring the pot against science. What is the solution to this? The framing seems way too simplistic to ever be useful. It makes it seem as though there was just the Good side that had all the answers all along, and the Bad side who just needed to do what the Good side said. In reality, it all came down to "who do you trust to both have yo…

There is no solution to this other than massive investment in education. And that isn't going to happen in our current world.

Re: Electronic Structure of LK-99

#413

Post by fragmede »

jacquesm wrote: post #412
ifyoubuildit wrote: post #410

Earlier quoted context omitted.

> Who did let us down: the people that were making things worse from day #1 by stirring the pot against science. What is the solution to this? The framing seems way too simplistic to ever be useful. It makes it seem as though there was just the Good side that had all the answers all along, and the Bad side who just needed to do what the Good side said. In reality, it all came down to "who do you trust to both have yo…

There is no solution to this other than massive investment in education. And that isn't going to happen in our current world.

Country? Yes. World? I'm not so sure.

Re: Electronic Structure of LK-99

#414

Post by feoren »

naasking wrote: post #325
feoren wrote: post #276

Earlier quoted context omitted.

From my perspective, LK-99 is by far the most likely to be an actual serious technological breakthrough. If it's real, it would be a breakthrough on the level of the discovery of electricity, the lens, the transistor, or the laser. One of those era-defining bottlenecks in the Civilization technology tree. GPT-4 is a nice iterative improvement over previous work, and a culmination of decades of research. It's not anyw…

> It's not anywhere near an AGI Conjecture. We literally have no idea how far away from AGI we are, that's one of the dangers.

Ehh ... we kinda do. However sure you are that calculators aren't close to AGI, I'm only slightly less sure that one-word-at-a-time auto-completion networks aren't close either. Both can do things the other can't. LLMs are not strictly more powerful than a calculator. They cannot add two numbers together reliably.

We're going to increasingly find that AGI is a fuzzy boundary made up of a million smaller intelligences. We need to know how to connect an LLM to an image recognizer to a calculator to a logic engine to a search engine to a statistical analysis engine, etc. etc. If you're looking for the actual AGI breakthrough, look out for some qualitatively new and interesting way to connect these brains together.

Re: Electronic Structure of LK-99

#415

Post by feoren »

marcosdumay wrote: post #376
taneq wrote: post #303

Earlier quoted context omitted.

> close to the limit of what we can accomplish with our current understanding of AI and our current availability of good data Isn't this literally the definition of 'state of the art'? It's the best we can do with our current understanding and data. This doesn't seem like a convincing argument that current ML techniques are tapped out.

No, what the GP is claiming is a theoretical end of line, where we would need some very innovative breakthrough to go further. Those tend to happen decades apart in a single field. And often the breakthroughs even come before the end of line does, so we jump a few decades. Now, the GP didn't put out the reasoning why he thinks so. Without further context, there's nothing there requiring you to agree.

Well said. I'd add that Sam Altman, CEO of OpenAI, also holds the same opinion. He has ulterior motives for everything he says, of course.

Every incremental improvement in utility of these tools requires exponentially more parameters and training data. Compare GPT4's 1.76 trillion params to GPT3.5's 175 million; nearly exactly 10x more. It's reasonable to say that getting the same relative improvement again will require another 10x parameters, and about 10x as much training data. GPT4 supposedly cost something like $20 million to train, so we're talking $200 million. We're talking entire data centers here. And the bigger problem is that GPT4 was already basically trained on the entire Internet (as far as we know). New information is being created all the time, but it will take many years -- decades, even -- before we have 10x as much useful information on the internet. And more and more of that information is regurgitated hallucinations from GPT4 itself and similar.

I'm not saying AI is ending -- there are lots of other avenues to explore. But I am saying we're hitting the limit of "just make it bigger" for LLMs.

Re: Electronic Structure of LK-99

#416

Post by feoren »

taneq wrote: post #303
feoren wrote: post #276

Earlier quoted context omitted.

From my perspective, LK-99 is by far the most likely to be an actual serious technological breakthrough. If it's real, it would be a breakthrough on the level of the discovery of electricity, the lens, the transistor, or the laser. One of those era-defining bottlenecks in the Civilization technology tree. GPT-4 is a nice iterative improvement over previous work, and a culmination of decades of research. It's not anyw…

> close to the limit of what we can accomplish with our current understanding of AI and our current availability of good data Isn't this literally the definition of 'state of the art'? It's the best we can do with our current understanding and data. This doesn't seem like a convincing argument that current ML techniques are tapped out.

> This doesn't seem like a convincing argument that current ML techniques are tapped out.

Absolutely right, I wasn't very clear -- I don't believe "current ML techniques" are tapped out. I do believe that we're not going to see a GPT5 for several more years, and a GPT6 for another decade+ (or if we do, it will be branding, not anything significantly better). We're at the top of the sigmoid curve for the benefit of just making bigger and bigger LLMs.

Re: Electronic Structure of LK-99

#417

Post by yetihehe »

sterlind wrote: post #337
yetihehe wrote: post #287

Earlier quoted context omitted.

> So did you make the engine and notice it's efficiency, or did you find it using first principles where no one else did? I found it from first principles, I still don't know why no one else thought about it, but it requires sort of splitting and reversing a normal stirling engine inside-out. Since then I've found several attempts at going in that directions which I took, but no one connected all those pieces yet as…

why can't that someone be you? if you've saved up enough for a home with a garage, can't you further save another $10k?

Right now that money would just mean that I can drop second job and do it in a month. Because I can't drop it currently, prototype will just take probably until the end of year, maybe a little later. But I already have everything needed gathered (electronic equipment and "ECU" parts, most of the required materials and some of 3d printed parts cad designs already done), now I only need time to manufacture what needs to be done pretty much manually. I don't need any more money for actual finishing, but thanks for suggestions.

Re: Electronic Structure of LK-99

#418

Post by yetihehe »

r2_pilot wrote: post #341
yetihehe wrote: post #287

Earlier quoted context omitted.

> So did you make the engine and notice it's efficiency, or did you find it using first principles where no one else did? I found it from first principles, I still don't know why no one else thought about it, but it requires sort of splitting and reversing a normal stirling engine inside-out. Since then I've found several attempts at going in that directions which I took, but no one connected all those pieces yet as…

I suggest designing it in CAD and possibly sending it to be metal 3d-printed, which would be cheaper than a whole garage(I feel your pain on not having space... my carport is currently saturated with my business's equipment and I'd really like to build a proper workshop). In addition to less metal shavings around, you'll have access to different geometry which may help your design.

This prototype won't require any 3d-printed metal for now, mostly 3d-printed plastic and some easily made at home metal parts. I already have some cad designs, I have a garage AND dedicated workshop and enough equipment and parts to finish this, but only thing lacking currently is time (I need two jobs until at least end of this year, then it will be better). But thanks for suggestions.

I don't want to whine or excuse myself, I only give an example of why some inventions take a long time to mature into working design. I will make this, just not in a month of real time.

Re: Electronic Structure of LK-99

#419

Post by ceejayoz »

postalrat wrote: post #409
mannerheim wrote: post #406

Earlier quoted context omitted.

1) There's a range of speeds that are possible depending on what you assume the turning angle of the jet was. That range doesn't go anywhere close to the speed of sound. 2) The burden of proof is on those who make an extraordinary claim. There is a perfectly mundane explanation of the cause of apparently fast-moving object which is supported by available evidence. Even if that depends on some assumptions, you would h…

A perfectly mundane explanation isn't proof. There isn't enough evidence in this case to prove it one way or another. That should be acceptable but for many people it isn't.

In that situation, though, https://en.wikipedia.org/wiki/Russell%27s_teapot applies, as we can’t prove the negative.

Re: Electronic Structure of LK-99

#420

Post by kridsdale3 »

thefurdrake wrote: post #384
jiggawatts wrote: post #212

Earlier quoted context omitted.

In an episode of Star Trek: The Next Generation a redshirt used a tricorder to check the weather and said they would have to beam up soon because a storm was approaching. At the time I thought to myself: “Bah, that’s never happening, it’s pure fantasy!” A few years later I told some friends we’d have to pack up our picnic because according to the weather radar app on my phone we had about 7 minutes until we got drenc…

We basically have tricorders, especially if you consider all the fancy things you can attach to them, like flir IR cameras. I think we've almost EXCEEDED star trek in that area.

I want a programmable hypospray. I'd be real popular at parties.
Post reply on HN