“Albert Chang claims that Large Language Models (LLMs) are currently poor at playing chess because they are optimized for pattern recognition rather than the deep, specific calculations required, often leading them to hallucinate moves.”