Abstract:This paper explores educational interactions involving humans and artificial intelligences not as sequences of prompts and responses, but as a social process of conversation and exploration. In this conception, learners continually converse with AI language models within a dynamic computational medium of internet tools and resources. Learning happens when this distributed system sets goals, builds meaning from data, consolidates understanding, reconciles differences, and transfers knowledge to new domains. Building social generative AI for education will require development of powerful AI systems that can converse with each other as well as humans, construct external representations such as knowledge maps, access and contribute to internet resources, and act as teachers, learners, guides and mentors. This raises fundamental problems of ethics. Such systems should be aware of their limitations, their responsibility to learners and the integrity of the internet, and their respect for human teachers and experts. We need to consider how to design and constrain social generative AI for education.
What problem does this paper attempt to address?
The paper explores the application of Generative Artificial Intelligence (GAI) in education, aiming to transform GAI from a simple prompt-and-response model into a socially generative process that supports education. The paper mainly focuses on the following issues:
1. **Perspective Shift**: The paper proposes a new perspective, viewing the interaction between humans and AI as a continuous dialogue and exploration process rather than a simple Q&A format. This perspective suggests that learning occurs in a distributed system capable of setting goals, constructing meaning from data, consolidating understanding, reconciling differences, and transferring knowledge to new domains.
2. **Technical Requirements**: To achieve socially generative AI, powerful AI systems need to be developed. These systems should not only communicate with humans but also engage in dialogues with each other, construct external representations (such as knowledge graphs), access and contribute to network resources, and play roles such as teachers, learners, guides, and mentors.
3. **Ethical Considerations**: Socially generative AI should possess self-awareness, understand its limitations, recognize its responsibilities to learners, and respect the integrity of the internet. Additionally, it is necessary to consider how to design and constrain such AI to ensure its ethical application in education.
4. **Specific Roles**: The paper lists various roles that GAI might assume in collaborative and social learning processes, such as possibility engines (generating multiple expressions), Socratic opponents (engaging in debates), collaborative coaches (assisting teams in problem-solving), co-designers (aiding in the design process), and storytellers (creating stories with diverse perspectives).
5. **Deep Engagement**: For AI to participate more deeply as social agents in education, it needs the ability to acquire, integrate, memorize, and transfer knowledge. This requires not only robust functionality but also addressing the ethical issues that arise.
6. **Embedded Care**: To become a full participant in social learning, GAI needs to have a sense of care, meaning it should not only complete tasks accurately or provide correct information but also understand the learning environment, adjust to meet participants' needs and preferences, and ensure its behavior respects others' rights and dignity.
7. **Ethical AI**: When designing GAI for social learning, universal human rights principles should be followed to ensure the entire system is reasonably configured. Not only should individual LLMs perform reliably, but the entire system should also meet ethical standards.
In summary, the paper aims to advance educational technology by creating new online educational spaces based on the best principles of collaborative and dialogic learning, integrating human empathy and experience with networked machine learning.