AI image recognition detects bubble-like structures in the universe

https://phys.org/news/2025-03-ai-image-recognition-universe.html

by PaulHoule • 1 year ago

110 78 1 year ago

karim79• 1 year ago

I truly hope that the common theme of the likes of "JWST Just Found Something Which Should Not Exist" etc will not be augmented by stuff like "we used AI(tm) to figure out X, Y, Z".

The last thing we need is hallucinations fucking up the more grounded astrophysics. I'm not saying that is what is happening, I just worry about stuff like this. AI causing us to bark up the wrong tree, and so forth.

m3kw9• 1 year ago

Wouldn’t past any scrutiny if they say AI enhanced the picture and found something new

Kye3• 1 year ago

Machine learning (AI) is used everywhere in astronomy. That's how they made the black hole image. Don't confuse the broader 60+ year old world of ML with transformers and diffusion models.

vlovich1232• 1 year ago

Based on the paper I linked, it seems like a straight up classical sampling and clustering with baysian hyperparameter tuning. This is “everything is now AI” slop that’s infected all grants, academic and private industry fundraising. There’s no neural net or LLM involved.

cma• 1 year ago

Not sure if there was an update/response to this but:

     1st image of our Milky Way's black hole may be inaccurate, scientists say

https://www.space.com/the-universe/black-holes/1st-image-of-...

dylan6046• 1 year ago

If anything, it's just going to call out a thing in the image that humans can then go and look at. Nothing in astronomy is ever "decided" by a single report. It gets looked at and scrutinized, and then committee style decisions are made about it. So if someone is using some ML to scan every image taken by JWST and calls out 1 cool thing for every other 9 things it finds that's "yeah, we know about that", then that's still quite a lot of new cool things. it'll just be able to do this faster and potentially much more in-depth than a human scanning across the images manually

brewtide2• 1 year ago

Yeah but what if we start seeing only using this new awesome tool? What if that becomes the new seeing apparati? THIS is the tool that breaks that mold? The tool that (near?)every field is also going to be considering to be the tool that's off limits, or be 'constrained?'.

What if we had that view with microscopes, back when?

I see the point being made above fully. If ai takes over it's because we are every day it seems like slowly placing that faith.

It's our wow. It's the future generations taken for granted.

"Much more in-depth" ways now just "the way".

wg02• 1 year ago

Yeah. Thanks for saying this. Please let be the real sciences real that have propelled the humanity forward with painstakingly detailed analysis by peer reviews and what not.

Let's keep AI for vibe coding, cat images and memes etc.

NitpickLawyer• 1 year ago

The last thing we need is hallucinations fucking up the more grounded astrophysics.

You're thinking of the wrong ML. Generative models "hallucinate" and it's as much a feature as it's a bug. ML in astrophysics is not generative. They use it for flagging, "binning" data and in general (simplified) classification.

ldjkfkdsjnv• 1 year ago

Eventually youll give in to the fact that ai is useful, and maybe revolutionary. Until then, continue using swear words and sticking your head in the sand

neuroelectron• 1 year ago

Clearly DSE (Death Star Events)

vlovich1232• 1 year ago

Can’t believe they don’t link to the actual paper: https://academic.oup.com/pasj/advance-article/doi/10.1093/pa...

Kye3• 1 year ago

> More information: Shimpei Nishimoto et al, Infrared Bubble Recognition in the Milky Way and Beyond Using Deep Learning, Publications of the Astronomical Society of Japan (2025). DOI: 10.1093/pasj/psaf008

It links to a doi.org URL which directs the browser to what you linked.

shagie• 1 year ago

And has the value of "it doesn't go dead as easily" - https://en.wikipedia.org/wiki/Digital_object_identifier

The DOI for a document remains fixed over the lifetime of the document, whereas its location and other metadata may change. Referring to an online document by its DOI should provide a more stable link than directly using its URL. But if its URL changes, the publisher must update the metadata for the DOI to maintain the link to the URL. It is the publisher's responsibility to update the DOI database. If they fail to do so, the DOI resolves to a dead link, leaving the DOI useless.

More about it at Digital Object Identifier (DOI) Under the Context of Research Data Librarianship - https://doi.org/10.7191%2Fjeslib.2021.1180

tokai3• 1 year ago

Factually wrong comment at the top. Very HN.

phito• 1 year ago

I can't see the link either.

aneidon• 1 year ago

I couldn't see their link either until I turned off Ublock Origin on the page

dylan6046• 1 year ago

what kind of nonsense are they doing with a link that uBO would block it? is it a 3rd party JS library that assembles an element that then places the link as stylized embed? I'd have expected more tracking type of stuff with it, but inspecting the element appears that the link is clean. my uBO did not block it????

Mistletoe2• 1 year ago

I've been trying UBO lite and giving it a chance, but have been having nothing but weird issues with it. The internet doesn't work as well anymore, so thanks Google. I think it's time for me to mosey on over to Brave from Chrome finally.

saint_yossarian• 1 year ago

It's blocked by "EasyList – Newsletter Notices", the annoyances lists are often a bit too aggressive.

__MatrixMan__5• 1 year ago

Not linking to the actual paper is exactly the sort of thing that I've come to expect of phys.org

Ruq• 1 year ago

My takeaway is that the universe is soda.

TZubiri• 1 year ago

It was poured 14B years ago according to fizz buzz theory

spacecadet• 1 year ago

Pour me some quantum foam.

quantadev• 1 year ago

These are probably mostly supernova remnants, but also, if you believe the "inside" (i.e. other side) of Black Holes are White Holes, it would make perfect sense to see White Holes in our universe where stuff is sort of "falling into" our universe and exploding outward. There might be an actual hierarchy of these kinds of things so that what we think was our "Big Bang" was actually the location where a White Hole emerged thru which flowed everything in our universe.

anigbrowl4• 1 year ago

You wanna hear my evidence-free cosmic structure theory? Of course you do.

If you shine a laser through a mass of soap bubbles it will unsurprisingly split into lots of smaller beams due to a mix of refraction and reflection. I have long held the suspicion that there's an isomorphism between gravitational and surface tension structures, that the multiplicity and distance of galaxies may be somewhat illusory, and that many of them are translated/rotated reflections of nearer ones. Laugh now, perhaps gasp in wonder later.

idiotsecant2• 1 year ago

What you're describing is gravitational lensing. It can make one galaxy appear to be several in different places or shapes. It is, however, well understood.

anigbrowl4• 1 year ago

I know what gravitational lensing is, but that's not what I have in mind (or rather, my gut - while I have a strong hunch about this, I do not want to invest the years of hard study to validate it or more likely end up in a dead end).

My hunch is that rather than space being a contiguous void with isolated mass of gravity behaving like tiny monopolar magnets, at the intersection between different mass systems there are 'surfaces' of some sort like the walls of a bubble in a pile of foam, and that if you could encounter this 'surface' you would either be repelled by it (most likely) or make contact and be able to slide around on it, and then once you got to the angles where walls joins, you would be able to zip along the intersections at great speed in ways that defy conventional physics. I can't really explain it in greater depth, it's an intuition that's half lifelong fascination with looking at soap films and what foam does, and half 'it came to me in a dream.'

btouellette• 1 year ago

There was a somewhat similar search for these duplicate galaxies as evidence for a universe with positive curvature. Because in that case if you look deep enough you'll see more images of the same galaxies although they'll be further back in time and possibly shifted in the way you're describing by the cosmic structure. It didn't pan out obviously.

wg02• 1 year ago

I don't laugh but it is an interesting idea. Most of the theoretical physics starts that way and then gradually verifying such assumptions with great care and experimentation over multiple generations of scientists.

zeckalpha2• 1 year ago

Sounds both like quantum foam and not at all at the same time

Danmctree• 1 year ago

You would see many more distorted galaxies if this kind of effect would contribute a lot of illusory galaxies

Udo• 1 year ago

What you're describing sounds like the curvature or topology of space would be non-flat. AFAIK this hasn't been completely ruled out, but so far every piece of evidence suggests the universe is flat over vast distances.

Intuitively I'd say if there was curvature or topological irregularities at the furthest distances we can observe, there wouldn't be a consistent redshift observed on far objects because some of them would be coming towards us instead of pulling away.

itishappy2• 1 year ago

I have long held the suspicion that there's an isomorphism between gravitational and surface tension structures...

Sounds like domain walls.

https://en.wikipedia.org/wiki/Domain_wall

anigbrowl4• 1 year ago

This concept is a bit too advanced for me (or the page is too minimal to easily understand), but it sounds fascinating. I'll read up more on it, thanks.

krige• 1 year ago

Remember when everyone suddenly started seeing channels on Mars?

jharohit• 1 year ago

I had eyeballed one in a random image from Hubble few years ago! Finally found my answer of what it was

https://x.com/jharohit/status/1479100020049678339?s=46

Great use of AI!

dylan6046• 1 year ago

There's the assumption, at least by me in the past, that every image ever taken by any telescope has been poured over that nothing new could be found by someone like me looking at it. It wasn't until I realized that most images are looked at by the people capturing the image while they look at the image for the one thing they were trying to study. In a Hubble/JWST type image, that point of interest might be < 10% of the captured data. (Think of all of those images of new discoveries that have been so zoomed in that it's nothing but a bunch of pixels) Once they finish with it, it just becomes part of the archive. There have been lots of discoveries of people combing over the archives to find things in existing data without ever needing any 'scope time of their own.

vintagedave• 1 year ago

'Bubbles' could imply something like space-time changes, and in face a couple of comments speculatively / hopefully read them like this: [1], [2]

But they're (sadly?) much simpler: Spitzer bubbles 'are formed by radiation and winds from massive stars, which carve out holes within surrounding dust clouds.'

So really just the blast radius!

-- https://www.spitzer.caltech.edu/image/ssc2013-05a1-bubbles-w...

I would _love_ it if either of both these comments were true, by the way. Space-time can be boring and restrictive. What if...? I love the idea of bubbles reflecting a smaller universe and what it might hint about FTL, for example, and I live in hope that we'll find abberations and abnormalities like this.

[1] https://news.ycombinator.com/item?id=43552920

[2] https://news.ycombinator.com/item?id=43552713

__MatrixMan__5• 1 year ago

This notion appears in Death's End by Liu Cixin (sequel to the sequel to The Three Body problem).

marcellus23• 1 year ago

What notion?

flanked-evergl• 1 year ago

Could be relevant to Conformal cyclic cosmology

In 2010, Penrose and Vahe Gurzadyan published a preprint of a paper claiming that observations of the cosmic microwave background (CMB) made by the Wilkinson Microwave Anisotropy Probe (WMAP) and the BOOMERanG experiment contained an excess of concentric circles compared to simulations based on the standard Lambda-CDM model of cosmology, quoting a 6-sigma significance of the result.

https://en.wikipedia.org/wiki/Conformal_cyclic_cosmology

AI image recognition detects bubble-like structures in the universe

Related Stories