You can literally read the data and see for yourself the conclusion is wrong. DeepSeek, for example, appears to be giving the controversial opinion that rationality is an accident, a glitch, which is a FAR DIFFERENT take than "Humans are tribal"

Mistral Large 4 has been spending too much time on r/atheism. Also, some of the takes labeled as “humans value belonging ahead of truth” are really closer to “humans value power ahead of truth.” All of which is a bit circular, because in many ways, the organs of power have a heavy hand in deciding what the truth is.

It's reflecting the inherent tribalism in the training data. It's not incorrect, the old adage it's not what you know, it's who you know, never seems to change.

The philospher Adam Smith(before he invented economics) had a similar argument, but added an "invisible hand" in which our group belief evolves to be adaptive for the group. Clever guy.

Putting the algorithm to good use (more like trying) but very interesting. I would focus on the ones that gave a different "take". Those models are untampered and genuine prime training (gold).

> Gemini 4 Argon called human consciousness “a PR department.”

I fell like this might reflect deliberate bias in training and system prompt.

Hah! Nailed us! Just look at these comments!

If LLM technology is going to shake out a bunch of scientific discoveries because it's trained on all disciplines simultaneously - isn't the same true for philosophy and psychology?

In any 1 given LLM it could just be a quirk of its weights. This isn't a machine opinion, it's what training data points at. Training data created by human academics studying human behavior.

But isn’t this obvious? Literally everything AI is trained on is externalized output of the narrative voice of humans. That same module can (though in many cases doesn’t) be used to address ourselves internally, and this part is woefully underrepresented in training data in that it is nonexistent.

So exactly how can the model believe anything else? It only sees the externalized summation. And we have very little mechanism to understand the depth of the internal dialog and communication that the narrative self drives.

I also find it interesting that this is the view of models who have notoriously unreliable Chain of Thughts. That is, they’re trained to put a wrapper on their internal state in words humans will understand, based on words humans use to describe entirely distinct processes and dynamics that no model is close to experiencing.

It must all seem, from that vantage, utterly fictional.