Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

What awful AI-generated illustrations.

Anyway, is there any justification for treating spacetime as a smooth manifold, except for mathematical tractability?



Yes. GR does it and GR gives good descriptions for a lot of observed phenomena.


Except, you know, all the elementary particles. I have this intrusive thought those are just bits of non-smoothly kinked/knotted spacetime but have neither the physics background or sense where to even start to look further into that.


Maybe what you're looking for is a "geon"?

https://en.wikipedia.org/wiki/Geon_(physics)


>held together in a confined region by the gravitational attraction of its own field energy

Hm, not quite, they'd be held together topologically but the https://en.wikipedia.org/wiki/Geometrodynamics is interesting.


>> Yes. GR does it and GR gives good descriptions for a lot of observed phenomena.

>Except, you know, all the elementary particles.

except, you know, all the elementary particles don't do a good job of explaining what GR explains


Hardly an expert, but it seems more likely to me that GR is the smoothed-out limit of a lot of much smaller stuff happening, than QM is somehow low-level wrinkles in an otherwise perfectly smooth spacetime.

I don't think a smooth anything is physically tractable. You need infinite resolution for perfect smoothness - essentially an infinite amount of information at every point.

Except there are no points. So somehow you have equations that define curvature floating in some kind of metaphysical space which somehow gets mapped to observable phenomena.

And if it's a noisy smoothness - where does the noise come from?


They're not AI generated, there's literally artist credits in the picture captions.


Perhaps they're post-edited somewhat but there isn't a way you're convincing me those are fully handmade. Look at the nonsensically warped tiles. Look at the rods suspending some of the globes but not others in a random fashion, look at the illegible "writing".


I think you're right. I don't really have a problem with AI generated art being used in articles like this. After all, what kind of alternative visual can you put alongside an article about fundamental theories of nature? If AI art helps to reduce the cost of scientific journalism (without decreasing the quality) then I'm ok with it.

Having said that, to produce art that actually works is clearly not easy. If you've ever used Midjourney or DALL-E, you'll know it's a challenge to get the tool to output exactly what you have in mind. It's also clearly hard to produce images that physically realistic. If mismatched tiles and unrealistic perspectives are noticeable then they'll be a distraction and detract from the reader's experience.


Oof yeah, I just saw the article pics. I don't have an issue with AI assistance, but the final output here was just kind of incoherent and confusing. Like an unrelated stock photo. I understand that it is hard to control but that just means that the artist needed to do a better job to make something more coherent, more informative and less generic.

Unless yeah as you said, it's just too abstract to put anything coherent, in which case I guess it works (though the mass weighing image surely could have been better than some generic sci fi thing)?


But these add nothing to the article but confusion. That's not how doing interference works. That's not any sort of weighting device. Could have been a generic galaxy-brain stock art and I'd be less annoyed.


Yeah and the style has the default DALLE-3 style written all over them. Isaac Young may be credited but I think Isaac is using DALLE-3 and I hope they are aware they have hired a "prompt wizard". :)


I think they even put the prompt as the image caption?


Doesn't the image caption contain a prompt?

When i use it in DALL-E 3, I get similar images.

"The image depicts an experiment in which heavy particles(illustrated as the moon), cause an interference pattern (a quantum effect), while also bendingspacetime. The hanging pendulums depict the measurement of spacetime. The actual experiment istypically performed using Carbon-60, one of the largest known molecules. The UCL calculationindicates that the experiment should also be performed using higher density atoms such as gold. The other two images represent the two experiments proposed by the UCL group, both of whichconstrain any theory where spacetime is treated classically. One is the weighing of a mass, the otheris an interference experiment."


The illustrations looked like some sort of steampunk drawings from a fantasy book.


I'm more annoyed with the captions.

> … Carbon-60, one of the largest known molecules.


Eh, C₆₀, one of the largest molecules with observed quantum interference, close enough.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: