i just learned that Eotriceratops was actually smaller than Triceratops

[andrewsaurus rex]

Which was big news, at least to me. The explanation was that it had a relatively larger head, leading to the initial estimates of it being larger than Triceratops.  But now it is considered to have been a bit smaller, but with a bigger relative head size.

Can anyone verify this?

[Shane]

According to who? Where did you hear this?

There's barely any described material of Eotriceratops; my understanding is that other than the skull, the only known remains are some neck and upper vertebrae and some ribs, and most of the remains were badly crushed.

Given how little there is to go on, I think any estimates of it being larger or smaller than Triceratops would have to be speculative estimates taken with a huge grain of salt.

[andrewsaurus rex]

Well, the source was Google AI and i confirmed it a couple of times by rephrasing the question, which often gets different answers from AI but this instance it gave the same response and reasoning every time.  I was surprised by it...i thought maybe AI had scanned some paper or new research that is not easily found online yet?

[Shane]

Well, the source was Google AI and i confirmed it a couple of times by rephrasing the question, which often gets different answers from AI but this instance it gave the same response and reasoning every time.  I was surprised by it...i thought maybe AI had scanned some paper or new research that is not easily found online yet?

Do not rely on Google AI, or any GenAI LLM chatbot, to provide you with accurate information.

These AI LLMs are under no obligation to be correct in the information they provide. Their responses are entirely generated based on probability, and they are only as good as the data they are fed.

Also, if you ask them something and they don't have an answer one way or another, they will simply make one up. Again, it is entirely based on probability. Each word in the response is determined probabilistically based on how likely it is to follow the previous word, and that probability is determined by the data that the AI is fed. That data is basically just whatever is out there on the internet.

Google AI doesn't favor scientific peer reviewed papers, and is just as likely to give you a correct answer as it is to parrot some random unsourced info from an online forum, or simply make something up.

To illustrate this, I just asked Google if Eotriceratops was larger than Triceratops, and the AI response was:

No, Eotriceratops was not definitively proven to be larger than Triceratops, although it had a larger skull.

However, when I asked Google if Eotriceratops was SMALLER than Triceratops, here's what it says:

No, Eotriceratops was generally larger than Triceratops, not smaller.

As you can see, these two responses directly contradict each other and the answer is totally dependent on how I phrase the question.

Disregarding the AI response and looking at actual Google results, the only information I found about Eotriceratops being smaller than Triceratops is a random post in a discussion on Dinopedia that provides zero sources. This is likely part of what is influencing the AI response.

[crazy8wizard]

Even the skull is incomplete, so the size of that piece is also partially estimated. The larger size estimate comes from *sigh*...Greg Paul's Princeton Field Guide to Dinosaurs. So even with proper sourcing, take it with a grain of salt because Paul has a habit of publishing sloppy or inconsistent sizings in his books.

[Shane]

Even the skull is incomplete, so the size of that piece is also partially estimated. The larger size estimate comes from *sigh*...Greg Paul's Princeton Field Guide to Dinosaurs. So even with proper sourcing, take it with a grain of salt because Paul has a habit of publishing sloppy or inconsistent sizings in his books.

Wikipedia lists the 9.8 foot skull estimate and cites Wu as the source for that information. For total size it states 8.5-9m, citing Paul as well as a dinosaur encyclopedia by Thomas Holtz Jr from 2012.

The size estimate would then put it about right on par with Triceratops estimates, but with a potentially larger skull.

But since we have few remains and their condition is mostly poor, everything is pretty much estimation and speculation until more can be found and described.

[andrewsaurus rex]

Alright, i guess that clears that up for the moment.  I knew this was the place to come for clarification.  Thanks gang.

i was skeptical of the results, but when it said Eotriceratops had a relatively larger skull than Triceratops and that's why its size has been estimated to be larger than Triceratops, that seemed nuanced enough to look further and ask questions.

I have no trust in AI responses; i find they are all over the place and depend on the phrasing of the question, but they can often point you in the right direction or introduce names and information that can greatly hasten your search for an answer.  And some times they can be surprisingly accurate.

As recent example, i had a growth on one of my fingers, just below the nailbed that looked a bit like a wart but was red and the nail on that finger had grown with a significant groove in it.  It wasn't of concern but i was curious so i typed into AI "growth on finger causing a groove on my nail" and the response was that it  is likely a "myxoid cyst". Never heard of it. So i researched myxoid cysts and that's exactly what it is.  No big deal and they generally go away after a year or so.  If it hadn't been for AI it could have taken me ages to figure that out; instead i had the answer in seconds.  So after that and many other examples i've experienced, i've learned not to totally dismiss AI but to verify what it tells you.  Hence me coming here for the Dino Toy Forum's AI (Actual Intelligence). 

[Shane]

Alright, i guess that clears that up for the moment.  I knew this was the place to come for clarification.  Thanks gang.

i was skeptical of the results, but when it said Eotriceratops had a relatively larger skull than Triceratops and that's why its size has been estimated to be larger than Triceratops, that seemed nuanced enough to look further and ask questions.

I have no trust in AI responses; i find they are all over the place and depend on the phrasing of the question, but they can often point you in the right direction or introduce names and information that can greatly hasten your search for an answer.  And some times they can be surprisingly accurate.

As recent example, i had a growth on one of my fingers, just below the nailbed that looked a bit like a wart but was red and the nail on that finger had grown with a significant groove in it.  It wasn't of concern but i was curious so i typed into AI "growth on finger causing a groove on my nail" and the response was that it  is likely a "myxoid cyst". Never heard of it. So i researched myxoid cysts and that's exactly what it is.  No big deal and they generally go away after a year or so.  If it hadn't been for AI it could have taken me ages to figure that out; instead i had the answer in seconds.  So after that and many other examples i've experienced, i've learned not to totally dismiss AI but to verify what it tells you.  Hence me coming here for the Dino Toy Forum's AI (Actual Intelligence). 

AI can be a useful tool if approached properly, but many people have quickly come to overly rely on it. It's important to remember that it's really only mimicking what it thinks sounds like a good answer, and is not actually capable of doing any actual research or vetting of the responses it provides. So while it can potentially provide correct information if the right answer is high enough in its probability data stream, for certain queries, especially ones that are more obscure or that there's less available information (or in some cases, too much information that isn't consistent), it's very likely to spout out nonsense responses, or just flat out wrong info.

Glad to hear you're using a skeptical approach, there's a lot of scary info out there about how much people are coming to rely on LLMs to give them immediate info, without really understanding the limitations or methods behind how they work.