As artificial intelligence (AI), including generative AI (GenAI), merges with Extended Reality (XR), we find ourselves at a crossroads in education. AI-XR metaverses can remarkably enrich teaching and learning, providing immersive and interactive experiences. Yet, these advancements carry the risk of reducing education to a mere marketable service, deepening social divides, and leading to complex ethical quandaries. This challenging dual-nature of AI-XR educational metaverses calls for a nuanced, Critical AI perspective. Educational AI-XR platforms, with their capability for covert monitoring, raise alarms about the privacy and autonomy of students and about increased surveillance and potential abuse of power. Moreover, the repercussions of XR, which can range from physical injuries to mental exhaustion, demand close examination for their possible adverse effects on student welfare. In the extreme, the line between simulation and reality may blur, possibly leading to a detachment from the real world, with students finding simulated environments more appealing than their actual lives. Further, the neglect of the physical self and environment in favor of virtual spaces poses significant risks to the holistic development of students, underlining the need for balance and integration of physical activity and real-world interactions within AI-XR curriculums. Safety and security issues are paramount, with the metaverse presenting new vectors for harassment and cyber threats that could disrupt the safe learning environments crucial for education. Educational metaverses must also cater to the process of meaning-making and alignment with the core values and purpose of education, which is to develop the personal, social, ethical and professional competencies of learners. We scrutinize the application of AI-XR in edtech, and highlight how mindless uncritical adoption of these technologies can devalue the human elements of teaching, reduce learner motivation, infringe on educational autonomy, and erode vital societal competences. This paper argues for directing technological advancements to augment, not undermine, social and personal welfare, while managing the inherent risks in AI-XR educational metaverses. There is a need for these AI-XR advancements serve to foster critical thinking, empathy, and ethical reasoning, rather than undermining them. Our paper proposes strategies to manage these risks within educational metaverses, advocating for policy development, ethical guidelines, and educational practices that ensure AI-XR technologies enrich rather than impoverish the educational landscape.

Keywords: Artificial intelligence, extended reality, educational metaverses, metaverse.