Anthropic engineer explains why Claude's writing got worse although the model got smarter

An Anthropic engineer noted that recent model updates prioritized technical performance in coding and mathematics over natural prose, leading to a denser and less conversational writing style. While these adjustments improved reasoning capabilities, they inadvertently made the AI's output feel less human-readable. The company is now working to address this shift in tone, acknowledging that prioritizing technical benchmarks can negatively impact the model's stylistic quality for standard users.