@404.media: “If LLMs Have Human-Like Attributes, Then So Does Age of Empires II,” is the title of Adrian de Wynter’s paper showing his work. He told 404 Media that absurdity can be a powerful tool. “I have this tendency to dial up things to 11 when I really think I need to make a point,” he said. “I should also note that absurdism is pretty standard in philosophy and theoretical computer science.” And so De Wynter built an LLM within AoEII using goats. “The point of the paper is to formally show that we anthropomorphise too readily, and that sometimes the claims we make with regards to LLM capabilities are too strong,” he told 404 Media. “It's not an easy task, given that ‘human-like attributes’ is a bit of an abstract term.” If you’re looking for human-like traits, you will tend to find them. De Wynter’s argument is that it’s possible to build a basic LLM within Age of Empires II that has many of the same internal traits of the chatbots people use everyday. The difference is the interface. When a person interacts with an LLM through the medium of AoEII and not a chat window the perception of human-like traits in the LLM vanishes even though the underlying tech is the same. @Evy Kwong Read at 404media.co
404 Media
Region: CA
Thursday 18 June 2026 17:28:21 GMT
Music
Download
Comments
JamesTanm :
unfortunately I think the only people who would understand this paper already understand LLMs enough to know they can't have feelings
2026-06-18 22:12:55
837
Fraize :
Reductio ad absurdum. This paper simply demonstrates a logic gate, not the training and data behind the LLM it drives. Computing power does not make it intelligent. Data and the organizational framework may, however.
2026-06-18 19:51:28
61
The_shakeweight :
this was one of my favorite papers I've ever read
2026-06-23 01:16:00
0
SerLava :
age of empires 2 IS a real person and I love him
2026-06-19 15:43:17
208
Tom from AI Basics :
You can explain how AI works all day, but until you explain how human feelings work differently, then you can’t say they’re different. Just be nice to the LLMs.
2026-06-22 18:54:58
3
Adam❌ :
By Anthropic’s logic, if I see an image of a person with anxiety in a mirror then the mirror is experiencing anxiety. 🤔
2026-06-19 03:57:48
107
Brosiden, King of the Brocean :
Ill believe AI is sentient when it calls in sick to work cause it wasnt feeling it
2026-06-20 00:44:18
13
aka.blitzbeat :
This misses the important point of emergent behaviors. A neuron or a perceptron are clearly not sentient but when taken at very large scales, sentient or conscious behavior can start to appear. Also worth noting there is no real definition of "consciousness" (philosophers have been debating this for thousands of years) so it's impossible to fully determine if an LLM is conscious
2026-06-19 12:54:59
15
James Palmer8623 :
I don’t think current LLMs are conscious but this is a bad argument given that consciousness is structural and doesn’t care about the substrate it’s running on. You could (slowly) simulate a human brain in any Turing-complete program like Minecraft or (yes) AOE2. We’re conscious meat and electricity!
2026-06-20 14:06:26
7
TP :
pointing at a tiny grain of sand and going "see, heaps are impossible, it's just sand"
2026-06-20 03:05:37
18
Ensign53 :
I agree with the outcome. but the example is a lot like saying "look, this light switch doesn't have feelings when it's flipped on and off, so humans don't have feelings because the underlying system is electric impulses in nerves"
2026-06-20 01:14:51
19
anplol :
if we had the compute to fully simulate human neurons would the same argument apply? i would claim that this is a spectrum phenomenon, not a categorical one, which means this is all silly (i dont terribly like anthropics anthropomorphization of claude tho)
2026-06-19 02:15:45
6
Sonder :
The amount of computing and power needed for “AI” which is just a glorified language model and is no where near complex enough to be a AGI, gen ai isn’t becoming sentient it’s being taught how to sound sentient so people like it more
2026-06-19 10:04:09
24
Viral Optimist :
LLMs are not conscious and effectively dead if nothing prompts them.
2026-06-20 16:56:20
6
Gracie_fan :
No one would say an individual neuron is human like but put a lot together and suddenly it is. So I’m not sure I agree but it’s a fun way to demonstrate the other side of the argument
2026-06-19 19:34:18
7
georggreasy :
AI is such a marketing job.
2026-06-19 12:30:40
17
burgahdestroyer :
how many goat gates to run DOOM
2026-06-19 05:58:58
8
Nalber3 :
I'm fed with AI bros talking about the "thinking process" of AI and that it reasons 🙄
2026-06-20 21:06:46
6
Clinton Ulfae :
Look I'm not making a claim on AI having emotions or not but you've shown the argument that because fly brains exist, humans can't feel emotion.
2026-06-18 23:14:35
16
Ardlboi :
this reduction stops working when you realise neurons are just moving ions around, how can that produce emotions?
2026-06-19 09:13:13
2
James :
a neuron might not look like it can be conscious but we know a lot of them together in complex configurations CAN produce consciousness. a cheap replication of an Ai doesnt look like it can produce consciousness but a trillion parameter models definitely seem curious about their own experience. not something a lot of things are capable of.
2026-06-20 02:27:29
4
vincentvandrey :
Emotions are just chemical reactions in our brains. Empathy is just pattern recognition of facial expressions and body language.
2026-06-19 20:15:33
1
Luke Kastner :
This is like saying humans don’t have consciousness because a handful of neurons in a petri dish don’t have consciousness.
2026-06-21 19:00:22
1
Alejandro Navarro :
1. Consciousness is an arbitrary human construct. 2. I’m NOT saying any LLM satisfies that construct. 3. If consciousness (as we define it) is the emergent property of a brain/nervous system responding to stimuli (in other words organized matter responding in predictable ways to some signal), then what makes consciousness (as we define it) categorically different than any thing we could build? We’re machines as much as a computer is, molecular biology is just engineering with really small machines, and that’s all we are. LLMs are missing features we have, like an innate understanding the physical objects we handle. But that’s just signal processing applied to a task we need for survival. It’s the same function applied to a different task, specialized in different ways. In the same way that if I define consciousness as the ability to comprehend binary, suddenly we’re out of the club
2026-06-20 20:14:14
2
Fernando M :
he made a simple computer in a game that uses a turing machine structure. this doesnt prove much.
2026-06-21 09:18:12
2
To see more videos from user @404.media, please go to the Tikwm
homepage.