Designing AI Technologies for Adult Learning: Guidelines and Best Practices

Malcolm Knowles' theory of andragogy posits that adult learners are independent, self-directed individuals with a wealth of life experience. Unlike children, who are often taught through pedagogy (a teacher-centered approach), adults require instructional strategies that are collaborative and directly relevant to their learning goals. Adult learning is often intrinsically motivated by immediate needs such as career advancement, credentialing, or reskilling. Furthermore, adults must navigate learning alongside employment, family, and social obligations, which significantly impacts when, where, and how they can engage with educational content. These contextual factors challenge traditional design assumptions that presume continuous participation or uniform learning goals. While andragogy highlights the unique characteristics of adult learners, foundational pedagogical principles like scaffolding, formative feedback, guided reflection, and structured activities remain valuable. However, these principles must be adapted to respect learner autonomy, leverage prior experience, and align with goal-oriented motivations. The Community of Inquiry (CoI) framework, comprising cognitive presence, teaching presence, and social presence, also offers a valuable lens for understanding meaningful online learning experiences, applicable to both K-12 and adult contexts.

This research draws upon the experiences and data generated from the National AI Institute for Adult Learning and Online Education (AI-ALOE) in the United States. AI-ALOE has focused on developing and deploying AI-powered learning technologies within diverse adult learning environments. Despite the varied nature of these deployments, stakeholders—including learners, instructors, administrators, and development teams—consistently identified similar needs, challenges, and design considerations. The recurrence of these patterns across different systems and contexts provided a strong impetus for a systematic, cross-technology synthesis of findings. By analyzing longitudinal deployment data, this study aims to extract actionable insights that can inform the creation of future AI-supported adult learning technologies.

Based on the reflexive thematic analysis of AI-ALOE deployments, 19 empirically grounded design guidelines were developed to inform the creation of AI-powered adult learning technologies. These guidelines are intended to address the unique needs and contexts of adult learners, moving beyond the limitations of systems designed for K-12 settings. While the specific details of all 19 guidelines are elaborated in the full paper, they collectively aim to enhance cognitive presence (learner engagement with content), teaching presence (instructor guidance and support), and social presence (learner interaction and collaboration). The guidelines provide practical recommendations for how AI technologies can be designed to be more effective, engaging, and supportive for adults pursuing education for career advancement, credentialing, or personal development, while acknowledging the constraints of their daily lives.

In addition to the design guidelines themselves and the heuristic evaluation method, this research introduces a novel 'Guideline Explorer' tool. This tool is designed to facilitate meaningful ideation and technology refinement by directly connecting the 19 design guidelines to the rich tapestry of stakeholder statements that surfaced during the analysis process. The Guideline Explorer acts as a bridge, allowing designers, educators, and developers to explore how specific guidelines relate to the expressed needs, challenges, and aspirations of adult learners and instructors. By providing this linkage, the tool empowers stakeholders to generate innovative ideas for new AI learning technologies or to enhance existing ones, ensuring that design decisions are grounded in the authentic experiences and requirements of the target user group. This fosters a more user-centered approach to AI in adult education.

While this study offers valuable insights and practical tools for designing AI in adult learning, certain limitations warrant acknowledgment. The research is primarily based on data from the AI-ALOE institute, which, while diverse, may not encompass the full spectrum of adult learning contexts globally. Future research could expand by investigating AI deployments in a wider range of institutional settings, industries, and geographical locations. Furthermore, the long-term impact and scalability of AI technologies designed using these guidelines require further empirical investigation. Longitudinal studies tracking learner outcomes, engagement levels, and the cost-effectiveness of these AI solutions would be beneficial. Additionally, exploring the ethical considerations and potential biases inherent in AI systems designed for adult learning, and developing guidelines to mitigate them, represents a critical area for future work. Continued refinement of the Guideline Explorer tool and its integration into design workflows would also be a valuable avenue for research.