Xirui Li

LLMs can exhibit surprising "strategic realism" when analyzing an ongoing geopolitical conflict, but their reasoning falters in politically ambiguous situations, revealing critical domain-specific limitations.

Despite high interaction density, AI agent societies like Moltbook fail to develop stable collective influence due to a lack of shared social memory, revealing that scale alone is insufficient for socialization.

RL's secret sauce for visual reasoning isn't better vision, but a targeted refinement of mid-to-late transformer layers that aligns vision with reasoning.