Approximately 970 million people worldwide suffer from mental disorders, yet over 60% do not seek professional help. Key barriers include low mental health literacy (MHL), insufficient awareness of psychological issues, and the overwhelming and non-transparent nature of information retrieval tools. Explainable artificial intelligence (XAI) can provide transparent and personalized feedback; however, its unique contribution to facilitating professional psychological help-seeking behavior remains unclear. Two randomized scenario-based experiments were conducted. Experiment 1 (n = 168) examined the effect of XAI on professional psychological help-seeking behavior. Experiment 2 (n = 178), utilizing a virtual simulation program based on AI search, tested the interaction between XAI and MHL and the mediating effect of psychological problem recognition. Both XAI and MHL significantly enhanced participants’ psychological problem recognition and professional psychological help-seeking behavior. A significant interaction was observed; XAI had a stronger facilitative effect on psychological problem recognition and professional psychological help-seeking behavior, particularly among participants with lower MHL. Furthermore, psychological problem recognition mediated the effects of both XAI and MHL on professional psychological help-seeking behavior. XAI can compensate for low MHL by reshaping individuals’ psychological problem recognition, thereby promoting the utilization of professional mental health services.

Adolescents’ internet adaptability (IA) is crucial for their online behavior and mental health. Serious games (SGs), as an emerging educational tool, hold promise for enhancing this adaptability through engaging, goal-oriented learning. Yet, direct evidence in this area remains limited. This viewpoint aims to clarify the mechanisms through which SGs enhance adolescents’ IA and to derive corresponding design principles that can inform educational practice and game development. Drawing on insights from both Chinese and international studies, this study adopts a cross-contextual perspective to explore how SGs can foster IA in varied educational environments. Beyond synthesizing existing findings, this viewpoint provides an integrated account of why IA is essential in contemporary digital life and how SGs can support its development. It proposes a 3-stage framework, illustrating how contextualized design, real-time feedback, and dynamic tasks promote experiential learning, self-regulation, and the transfer of online skills. Based on this framework, the study further articulates 6 core design principles: clear goal definition, interaction diversity, contextual authenticity, immediate, scaffolding and explanatory feedback, a dynamically adaptive learning environment, and safety-by-design for digital well-being. These principles translate the core characteristics and mechanisms of SGs into actionable guidance for developing effective IA interventions. By synthesizing theoretical insights with practical considerations, this viewpoint highlights how SGs can serve as accessible and scalable tools to support adolescents in navigating increasingly complex digital environments. Together, these insights provide practical implications for educators, curriculum designers, and digital game developers seeking to foster adolescents’ safe, responsible, and adaptive engagement in online environments.

Background: Large language models (LLMs) are increasingly being used for text-based emotional support. However, general-purpose LLMs may not consistently recognize emotional cues, adapt supportive strategies to contexts, or generate responses aligned with psychologically informed communication. The limited availability of large-scale, structured Chinese counseling data further constrains the development of models for context-sensitive emotional support. Objective: We developed and evaluated YuanChat, a family of specialized LLMs for context-sensitive emotional support; assessed whether E-URIAL improved response quality; and compared YuanChat with base LLMs. Methods: We constructed 2 datasets from anonymized real-world counseling sessions: ECD-3 (104,897 counselor response turns) for fine-tuning and D _gold (9413 manually annotated counselor responses). Annotations covered emotional support conversation stages (exploration, comfort, and action), counseling techniques (clarification, support, and challenge), and empathic effects. We developed 2 YuanChat variants using GLM-4-9B-Chat and Huozi 3.5 as backbone models using parameter-efficient fine-tuning with low-rank adaptation (LoRA). We then developed E-URIAL, an inference-time framework combining predicted emotion and support strategy labels with retrieved labeled exemplars. Given the dialogue history through the help seeker’s latest utterance, each model generated a counselor-style response. We compared YuanChat with GPT-4, ERNIE 4.0, CharacterGLM, GLM-4-9B-Chat, and Huozi 3.5. The evaluation comprised (1) automatic evaluation on in-domain and external datasets, assessing empathy identification and empathic expression; (2) evaluation of empathy-related situational reasoning using an expert-developed multiple-choice benchmark; (3) evaluation by clinical experts (n=8) of empathic attunement, support strategy appropriateness, perceived helpfulness, and safety and professional boundary adherence; and (4) a complementary evaluation by volunteer master’s students in clinical and counseling psychology (n=50) of response fluency, empathy, and perceived helpfulness. Results: Fine-tuning increased YuanChat-Huozi3.5’s empathic expression score from 0.57 to 0.68; empathy identification score increased from 0.54 to 0.59. In expert evaluation, YuanChat-Huozi3.5 received a higher empathic attunement score than the base Huozi3.5 model (1.279 vs 0.327; mean difference 0.952, 95% CI 0.746‐1.152). Using professional counselor responses as the reference, the mean differences for YuanChat-Huozi3.5 were 0.148, 0.079, 0.039, and 0.002 in empathic attunement, supportive appropriateness, perceived helpfulness, and safety and professional boundary adherence, respectively. No statistically significant differences were detected between YuanChat-Huozi3.5 and the counselor references (all Holm-adjusted P ≥.29). Higher scores were observed when E-URIAL was applied. When E-URIAL was applied with 3 retrieved exemplars, YuanChat-Huozi3.5 achieved an empathy identification score of 0.69 versus 0.59 without E-URIAL, an empathic expression score of 0.79 versus 0.68, and a perceived helpfulness score of 0.74 versus 0.61. Conclusions: The findings indicate that fine-tuning on real-world counseling data and annotating models with structured emotional support labels improved context-sensitive empathic responding in LLMs. Without updating model parameters, E-URIAL further improved empathy identification and empathic expression across multiple models. Together, these methods provide a scalable, theory-informed approach to enhancing the emotional support capabilities of LLMs.