Not So Fast: Mapping the Learning Speed and Sophistication in GenAI

Among many functionalities, Generative Artificial Intelligence (GenAI) can model the topology and semantics of user-supplied datasets - a functionality required to evaluate learning levels through mind maps. Since GenAI evolves by orchestrated changes to the underlying algorithms and, organically, by learning, we need to understand this evolution's speed and reliability. We conducted two experiments tasking ChatGPT with scoring mind maps drawn by 113 undergraduate students describing their motivation and deterrence towards entrepreneurship. Scoring used a five-dimensional model consisting of self-efficacy, internal locus of control, need for growth, intrinsic motivation, and resilience. We repeated the analysis on the original dataset after eight months to time the evolving pace and sophistication of the tools used. The results show that we should not fall into the “hype” curve typical of the beginning of any emerging technology. While the pace of learning in GenAI is unprecedented, caution is necessary when rechecking data and analytical techniques.