Co-Existence and the End of Co-Intelligence

This One Useful Thing article, titled "Co-Existence and the End of Co-Intelligence," centers on how Ethan Mollick positions a new phase that follows his previous book, Co-Intelligence. The core thread that can be read from the title and public information is this: Mollick argues that the common way of using AI during 2023 to 2024 was primarily humans collaborating back and forth with chat-style models through prompts—what he calls the co-intelligence mode. But as AI capabilities, agentic workflows, and degree of everyday embedding have increased, the question has shifted from "whether to use AI" or "how to work alongside AI" to "how humans coexist with AI once it becomes a routine intelligent agent in work, education, creation, and life." The new book, Co-Existence, continues in this direction, arguing that the future is not merely about replacement or assistance, but about humans and AI continuously interacting amid uneven capabilities, attribution of responsibility, authority over judgment, dependency risks, and the boundaries of creativity. For Taiwanese readers, this piece reads more like an opinion-laden publishing preview than a model release or technical tutorial; its significance lies in offering a framework for understanding society and work after AI has become widespread. Mollick also uses experimental material such as "how to pitch a book to an AI" to maintain his style of treating AI as something that can be tested and interacted with, but that still requires human judgment. The article does not present any specific new models, benchmarks, or tool features, so when interpreting it one should avoid treating it as product news; it is better viewed as commentary and a signal that AI adoption is entering its next phase.