Can AI produce writing that we want to read? (www.newyorker.com)

A recent exploration into AI-generated writing has raised significant questions about the future of literature and education. As instances of AI-created content become more prevalent, educators worry that students might rely on AI summaries instead of engaging with the writing process, potentially diminishing their learning experience. The critical concern is whether AI can produce compelling, human-like prose, given that current outputs often exhibit peculiarities—such as awkward metaphors and static characters—that betray their artificial origins. In a creative experiment, the AI model Claude was tested on its ability to mimic the styles of famous authors. While initial attempts revealed clear tells that signaled AI authorship, subsequent tweaks allowed Claude to produce more convincing passages that fooled many players in a real-time guessing game. Despite improvements, the generated content still often lacked narrative dynamism, reflecting characters who largely remained inactive. This limitation underscores both the potential and the current shortcomings of AI in literary production, sparking a conversation on the implications for both creative writing and the role of educators in a future where machine-generated text becomes increasingly indistinguishable from human writing.