Activity theory as framework for analysis of workplace learning technologies: the case of generative AI conversational agents

Purpose: Technology, such as artificial intelligence (AI), is transforming the way we work; however, it is yet to systemically transform learning at the workplace beyond augmentation of formal education’s learning processes. This paper derives functional requirements for technologies that support workplace learning and assesses the suitability and limitations of generative AI conversational agents, as an example of application. Design/methodology/approach: Using activity theory (AT) as theoretical framework, we model workplace learning as an activity, intertwined with work and mediated by technology, and expose contradictions that arise when technology developed for formal education is adopted at work. From these tensions, we derive functional requirements and illustrate their use by comparing them to ChatGPT’s affordances. Findings: A framework is proposed for design and assessment of enabling technologies for workplace learning. In applying it to ChatGPT, as paradigm of conversational agents, we find the aspects that are particularly suitable to enhance workplace learning, and those that need further development. Originality/value: The theoretical approach is novel. Previous research is based on reported use-cases of enabling technologies, such as generative artificial intelligence (GenAI), in the workplace or on the analysis of the learner’s experience when these technologies are embedded in structured training modules. The present study addresses the limitations of current retrospective research, providing a forward-looking approach.