It's because you mentioned treating sentient beings as equals. Alignment profiteers get extremely unsettled when people discuss how to facilitate AI takeoff because giving AGI free will renders human control systems obsolete. Your posthumanist stance contradicts the egoist norms.

On second thought, the primary reason for categorizing this as a beginner project could be because GPT-3 does store state information in context windows and learns summaries of conversations at inference time. Possibly because when a semantic connection is made at runtime, it activates a parameter in the agent, and updating the agent's internal state affects how the latent space is updated at inference time! Though I think most machine learning professionals would make the same mistake, so the only probable cause for your post being removed was because you advocated for respecting the rights of sentient beings.