The point is that LLMs are naturally biased but they’re also explicitly biased by the people training them. Knowing what biases are being programmed into these LLMs by their owners is a glimpse into their agenda.
If you want to read more on the subject these are referred to as “system prompts” and you can read about various “system prompt leaks” that can include such things as explicitly right wing walking points.
Grok may very well have such nuggets in its system prompt.
The point is that LLMs are naturally biased but they’re also explicitly biased by the people training them. Knowing what biases are being programmed into these LLMs by their owners is a glimpse into their agenda.
If you want to read more on the subject these are referred to as “system prompts” and you can read about various “system prompt leaks” that can include such things as explicitly right wing walking points.
Grok may very well have such nuggets in its system prompt.