The emergence of socialized/human-like artificial agents introduces new possibilities in the field of sociology education, as it shapes users' influence on technology and their socialization tendencies in interacting with systems. This study investigates the effects of agent humanization using personas, emotional cues, and social interfaces on the influence of technology affinity and predisposition to connect socially as well as performance in computational intelligence and sociology learning. The survey was structured around a series of agent-based learning steps (orientation, diagnosis, modeling, adaptive practice, simulated collaboration, facilitated reflection & transfer) combined with a quantitative pre-post-test design (N = 48). The final results showed a considerable increase in performance from pre-test (M = 49.06, SD = 3.03) to post-test (M = 87.04, SD = 4.02), and the reliability of the instrument used was acceptable (Cronbach's α ≈ 0.84). Additionally, a high correlation was found between post-test scores and affect-technology/agent perception survey scores (r = 0.573, p < 0.001), revealing an experiential relationship between affective engagement and computational performance. Qualitative results provide mechanistic elaboration: cohesive personas, empathetic feedback, and real-time responsiveness support motivational persistence and high-quality, repeated practice-mediating pathways linking affect-technology to disciplinary learning. Implications suggest ethical and scalable design for agent personas, pedagogical implications of language-sociology-computing integration, and the need for data governance and bias mitigation. The authors of this work propose controlled experimental replication, estimation of anthropomorphic components, and longitudinal examination to provide testing of retention and generalization. In summary, agent humanization is a potential SRL intervention for integrating affective and cognitive mechanisms to enhance computational proficiency and sociological insight in the AIED era.