The site in question was found earlier: https://bsky.app/profile/indyfromspace.bsky.social/post/3n22...
The Mexican govt may not have had the right records, but you can see what the site looked like at the time it was first found here: https://bsky.app/profile/indyfromspace.bsky.social/post/3n22...
Here is the referenced publication. Pages 13-14 show the structures. https://www.sciencedirect.com/science/article/pii/S2352409X2...
Did his Gemini AI agents just use Google search, find the publication, check the coords, and claim it as a discovery? Ha
This is a good example of the difference between an LLM providing a theorem and a person deducing theorem, and all the juicy knowledge we get from truly understanding things: the Google SWE found something he thought was undiscovered. He asked AI to make an image what it might've looked like and speculated that it was used by the local elite.
But Dr. Estrada-Belli found something that was undiscovered. Finding this structure wasn't his number one goal - it was the side effect of a larger anthropological assessment that allowed him to improve estimates about population density of the north versus south, how Mayan homes and farms are distributed, what this means about social relationships between the elite and common people, how different regions were interconnected, etc.
Interestingly, one of Dr. Estrada-Belli's primary conclusions is not that these structures were only used by "the local elite," but that the closeness of residences and farms to these community sites suggests that the elite and the common people actually had a much closer and nuanced relationship than we previously thought. That's the opposite conclusion you'd get from the Google SWE's assessment.
Knowing that a Mayan site exists is one thing. Understanding what it means is another.
Very reminiscent of my experience pointing persistent agents at computational physics problems. It can find amazing results and really surprise me, make a write up, and convince me it's a real discovery... until i tell it to do a literature search, and be meticulous. Then it comes back saying "prior art has reduced the novelty of the discovery", and finds papers explaining the discoveries very easily and intelligently.
Perhaps google SWE should have requested a literature search before publishing his results as novel
It’s like a shell game. It’s likely overfitting on its training data.
I enjoy the contrite rationalizations and turns of phrase LLMs use when encountering a discrepancy.
"Prior art has reduced the novelty of the discovery" is a great one
IMO, his primary mistake was not turning the agent swarm on a proper research task that exhaustively determined whether it had been discovered or not.
Well the Google SWE didn't make a mistake, the media made a mistake in misreporting it, and we at HN are making a mistake by giving it undue attention.
We are reading this article because it continues and amplifies the AI hype machine, and no other reason. If a motivated amateur had made the discovery through any other means, and motivated amateurs do things like this all the time, then it would never have gotten an article, never shown up on the HN feed.
The first thing I saw when I tried to follow this link is a captcha. So it's no wonder that the agent wasn't able to find this. We are siloing our knowledge unnecessarily.
If the "AI" scrapers honored HTTP 402 I'd be happy to let 'em in.
Or necessarily depending on your viewpoint.
Who says the agent was treated the same as you?
AI is already fully capable of solving the majority of/all captchas, you just aren't supposed to do it due to TOS.
I expect with time these kinds of things will change.
Lets take a moment to consider why that has started happening.
The lineage of every LLM of relevance today was the largest theft of intellectual property in human history. Every digitised creative work, whether in the public domain or not was taken, often outright pirated, with the explicit intention of being able to trivially produce compelling forgeries. Physical works are being enthusiastically destroyed to digitise them to further feed the hunger. And outside of a few court cases, the original creators across the world have been paid nothing for their work having been stolen and repurposed to put them out of a job.
Meanwhile, those same digitisation efforts have resulted in badly behaved scrapers from all comers absolutely thrashing the entire web, in many cases exponentially increasing the cost of hosting for smaller entities. Larger entities care less about that but are still enthusiastically walling their gardens so they can monetise the output of their communities rather than continuing to allow it to be stolen freely.
We are siloing our knowledge out of desperation. I'm sympathetic to the idea that knowledge should be free but that isn't possible when bad actors like Google and OpenAI are preying on the commons.
Theft/stolen implies it's been taken and the owner deprived of it. Pirated or copied is more appropriate, but still not quite exact as it's been encoded into what amounts to lossy memory. If I ask a person to tell me about a paywalled article, they'll share what they can remember and likely get some facts wrong. An LLM is the same (these days; GPT-3 was another story).
I do agree that the methods of obtaining much of the material (eg. the published works torrents) were stupidly wrong and they should pay for that. However, legally acquired media being "read" by the training process isn't so different than it being read by a voracious reader: it's lossily encoded in (what amounts to) memories.
The insane scraping levels and behaviors ... not wrong. If only there were companies who already did that responsibly they could contract with for data. Hmm. (But they have to "own" it, as is the SV fashion these days.)
From my perspective, the bad actors are the corporations and greedy people who turned copyright into a perpetual rent on IP. If copyright terms and IP laws were reasonable, I'd be 100% with you. But the idea that a novel written in 1940 is still not public domain, just so the heirs (or more likely, corporate purchasers of the rights) can continue to profit on a work that was itself but on the shoulders of thousands of years of civilization? Screw that.
The Walt Disneys of the world screwed themselves on this one. If copyright was still the 20+20 it started as, and if the bulk of people being "stolen" from (don't get me started on using theft to refer to IP), then I think most people would be much more up in arms over the AI industry.
But, in a world with sane copyright laws, perhaps AI companies would have different approaches. Today, if they want anything after 1930, they're either breaking laws or in uncharted territory.
And siloing our knowledge is a silly response even if you are a copyright maximalist.
Not we, just Elsevier. As long as Academia sells its research, information will not be free.
Clicked the GP link and I got a “this post is private” page. Clicked your link and got a Cloudflare page that hangs and I can’t move past. That’s the Internet in 2026
> I can’t move past
Come on, George, you're clearly a ramblin' wreck from Georgia Tech. You've taken nearly every course GT has offered over the past century, you can get through a captcha. Live up to your name.
To some extent it's still an interesting result: an independent validation of the original finding, and a proof of concept that this technique could make such a discovery.
IF the llm did not have the paper on it's training set. Which we do not know.
I hope he collaborates with archeologists to apply these methods to their work.
Care to share the source without a paywall?
"This author has chosen to make their posts visible only to people who are signed in."
>remote sensing is difficult and any archaeologist knows to triple check before saying something is "new"
>Francisco was very kind and shared his publication image with me
...can she post the pdf?
Seems likely the google SWE that wrote the blog wouldn't have claimed this was new if there was a public way to access this information. I understand archaeologists have valid reasons to keep location details private, but complaining the OP didn't "check" thoroughly (when she can't link to the data he should checked in order to validate it was new) reads more like a shortcoming in how this knowledge is shared than a failure of OP to check.
open access.
https://www.sciencedirect.com/science/article/pii/S2352409X2...
Not to agents
I told my agent to bypass Cloudflare captcha's and it happily did so.
This author has chosen to make their posts visible only to people who are signed in.
The post says:
> This story is going wild on X and elsewhere. The site was already published by Dr. Francisco Estrada-Belli of Tulane University ages ago. This is why we have peer review. Mexican gov didn't have it in their database, but that doesn't mean it is unknown/unmapped. https://pietromiotti.com/latest/maya-site/
Basically: already known, not in the Mexican Gov database, but in published literature... and open-access.
Link to the previous publication: https://doi.org/10.1016/j.jasrep.2025.105288 (Journal of Archaeological Science: Reports 66 (2025) 105288)
Can you please reply this on main thread so that it confirms that this overhyped non-peer reviewed vibecoded approach doesn't work
The approach did work, but the discovery isn't new. Independent re-discovery happens all the time science.
but the paper was probably in the AIs training set and they could have accessed it at any time during the run.
Yes but someone who knows what they are doing with years of experience i the field probably slows down, is more thorough, and goes through a peer review process before rushing to announce to the world their new discovery.
> Me: That's awesome! Please check which of these villages haven't been discovered yet. Search both the literature and the archives.
> Agent: One major village doesn't seem to have been registered.
The idea of just brute forcing discovery with agents on lidar is really cool, but doesn't the story ultimately rest on the agent making a mistake in its literature search?
Yes, and I think this point and the post you're replying to are the central fact of this story, which the OP should acknowledge.
My own attempts to use agents to develop new historical knowledge [1] point to this as being both a very promising and interesting domain, but also one where the agents reliably and repeatedly make claims of novelty that are based on having read ~80% of existing research. Obviously to make an actual well-grounded claim of novelty you basically need to approach 100%.
This is partly why math is so tractable for this sort of thing - mathematics papers tend to be publicly available and searchable, whereas for history, it might be the case that you have an agent swarm duplicating a find that some French person in the 1870s previously reported in an obscure undigitized book.
[1] Detailed more here: https://resobscura.substack.com/p/using-opus-55-to-discover-...
Did anyone in the replies link to the actual paper in which this site was mentioned? Because that's a weird omission if you are trying to argue that something was already in the literature.
you can see they're obviously the same structures in the images. just search for the author and you'll find the paper.
Then the title is misleading, yes? Should be something like "Google SWE uses AI to find a pyramid site that was already discovered".
Not news.
The moral of the story is that if you thought you found something that was undiscovered: no you didn’t.
BUT, if AI thinks it’s undiscovered then its discovery might be poorly documented, so if you were to go to the site by yourself you might be able to explore it on your own and raid tombs, etc, and no one will notice.
Tomb Raider is a video game.
In real life, that’s actually just vandalism of an archeological site for personal gain.
Potato, po-tah-to.
Don't ruin Indiana Jones for me.
even if it was discovered, I'd say most of the accolades go to NASA and LIDAR, the fact that AI agents were used to discover it isn't a big deal.
Now I'm very interested in seeing the transcript/session from what the agent actually did.
We already know since before that agents takes all sorts of possible shortcuts to find the answer they think the user was looking for. It doesn't seem impossible to me that the user asked something that made the agent feel OK with highlighting something that wasn't logged in the Mexican database as a novel discovery, even if it wasn't.
But again, transcripts/session logs would make it extremely clear exactly what has happened, and I'd take the author not publishing those as trying to hide something about how the agent actually approached this.
"I'd take the author not publishing those as trying to hide something about how the agent actually approached this."
That is an irrationally cynical take. Everything described in the post is clearly within the abilities of a frontier LLM. The previous discovery is obscure enough that the Bluesky author is not citing any actual publication but instead encouraging people to reach out to a scientist and receiving images through private communication. The reasonable interpretation is that the model stumbled upon a previous discovery in the manner described, and the author/LLM failed to find the previous discovery because it is not any published database. The author is not running a scam for some unclear benefit that would be immediately revealed when he published it.
> That is an irrationally cynical take. Everything described in the post is clearly within the abilities of a frontier LLM.
No, the burden of proof that it's actually novel sits on the author of this post, and we assume everything is cheated/shortcutted by default when it comes to agents and LLMs, unless we can prove it actually "found" this, not just looked up stuff that wasn't properly in the DB yet but already written about.
> The reasonable interpretation is that the model stumbled upon a previous discovery in the manner described
That'd be reasonable indeed. Instead the author chose the headline "My AI Agents Found an Unmapped Maya Site" and the first sentence is:
> By analyzing NASA lidar data, my AI agents discovered a Maya site in Mexico that, as far as anyone can tell, no one had reported before
Which without any evidence of what the agent actually did, I don't think we should just assume what the author says is true because again, we know that agents takes as many shortcuts as they can.
Plenty of times the people pushing LLMs are over-exaggerating more than in other ecosystems, and the community should rightly push back on these things, when they're not presented properly with proof and evidence. Otherwise we'd be doing previous actual archeologists a huge disservice.
Imagine being the archeologist who actually discovered this site, and waking up today reading some random person taking credit for your discovery! Kind of feels like people here often forget there are other people involved on the other side of things.
He checked with the Mexican government before claiming it was new, which is reasonable due diligence surely?!
So now this apparently amazing feat that was rushed into publication is actually just an out of date DB.
Peer review also protects the authors for making a fool of themselves. QED.
Looks like it was peer (or rather superior) reviewed. I don't think the author was trying to be malicious and just wanted to blog about something cool they found.
That is not how peer review works typically, but at least the guy did try to confirm with actual experts. So no, no evil here.
[I withdraw this comment.]
That's a pretty extraordinary claim.
I don't follow, what exactly are you saying? Are you claiming that a scientific article (shared in a sub-thread alongside yours) is fake?
You need a shave, here, use Occam's razor.