Abstract: ChatGPT is pioneering new frontiers in human-computer interaction and triggering significant transformations in the human-computer collaborative educational ecosystem. Based on Habermas's theory of communicative action, this paper critically analyzes the linguistic characteristics of ChatGPT, clarifies the "quasi-human comprehension" concealed beneath the language representation of intelligent technology, and uses this as a focal point to project the potential challenges and relief strategies of ChatGPT in transforming the human-computer symbiotic educational ecosystem. Under these potential challenges, the "parallel symbiosis" model of ChatGPT's transformation of the educational ecosystem is revealed. That is, educational agents, based on whether they are aware of the identity of their communication partners, derive a dual-track path that conforms to instrumental action and communicative action, and uses "human empathy" to empower "machine understanding" as a pivot to unblock the channel from instrumental action to communicative action. Meanwhile, instrumental action may successively trigger crises such as "isolation", "alienation", "value distortion", "moral weakening", and even "systemic imbalance" in the educational ecosystem. These challenges need to be addressed through strategies such as meeting the ideal conditions for human-computer symbiotic communication, constructing a multimodal open system, innovating the educational evaluation system, and improving technical ethical norms.