Comprehensive sexuality education in primary education: analyzing the impact of an educational intervention on knowledge, perceptions, and communication-a quasi-experimental study.

In the third decade of the 21st century, artificial intelligence (AI) has evolved from a peripheral technological concept into a transformative force that is reshaping human cognitive patterns, social practices, and cultural production. Its penetration into the arts has been particularly rapid and far-reaching, not only continually challenging the traditional philosophical foundations of creativity, authorship, and aesthetic value (; ), but also increasingly intervening at an unprecedented depth in the micro-level decision-making processes of artistic creation, performance, and education. Music, as an art form that is highly dependent on intuition, emotional resonance, and immediate interpersonal interaction, is undergoing a structural transformation driven by AI technologies. From AI composition, such as Google’s Magenta project, and intelligent arrangement to real-time accompaniment generation, these technological tools are increasingly moving beyond their auxiliary roles and becoming indispensable co-creators for musicians (; ; ; ; ). However, amid rapid technological advancement, a core question remains underexplored in empirical research. When AI evolves from a passive tool into an intelligent entity capable of offering professional opinions, how does it influence the development of confidence and decision-making strategies among human artists, particularly postgraduate students who are at a critical stage of professional development and whose professional identities are still being formed, in advanced collaborative settings? This study focuses on vocal coaching, a highly specialised field that relies on nuanced interpersonal judgement and deep trust, and aims to systematically examine the effects of AI as a Second Opinion on collaborative confidence and decision-making strategies among postgraduate vocal accompanists. Vocal coaching is far more than piano accompaniment; it is a high-level collaborative art form that integrates musical text analysis, multilingual vocal guidance, historical stylistic interpretation, psychological support, and real-time artistic co-creation (). In this process, the vocal coach must establish a relationship of trust with the singer, grounded in mutual respect and deep understanding, in order to jointly explore the emotional core, technical details, and aesthetic expression of the work. Every minute decision, whether concerning tempo, dynamics, phrasing, or timbre, arises from a complex negotiation between both parties, informed by their respective experience, intuition, cultural understanding, and immediate feedback. This decision-making process is characterised by uncertainty, subjectivity, and contextual dependence, which also constitute its distinctive artistic appeal. However, for postgraduate students who are transitioning towards becoming independent artists, this inherent ambiguity presents a significant learning challenge. Due to limited authoritative practical experience, they often experience deep self-doubt regarding their professional judgements, which in turn constrains their initiative, voice, and confidence within collaborative interactions (; ; ). Traditional teaching models largely rely on the tutor’s first opinion, which represents a unidirectional and authoritative mode of knowledge transmission. Although this approach provides clear guidance, it may also risk constraining students’ critical thinking, independent decision-making abilities, and the development of their individual artistic voice (; ; ). In recent years, the concept of an AI second opinion has gained increasing attention in high-risk and complex decision-making domains such as medical diagnosis and financial investment (; ; ). Its core premise is to introduce an independent intelligent system based on large-scale data and algorithmic models to complement, rather than replace, human expert judgement. This approach aims to provide diverse perspectives, reduce cognitive biases such as confirmation bias, broaden decision-makers’ perspectives, and ultimately enhance both the confidence and robustness of their decisions (; ; ). Extending this interdisciplinary paradigm to the arts, particularly to the highly subjective and complex collaborative context of vocal accompaniment, carries significant theoretical and practical implications. Contemporary AI systems, especially those based on deep learning for Music Information Retrieval (MIR) and generative modelling, are capable of conducting multidimensional and objective analyses of musical scores and audio signals with a level of speed and precision that exceeds human perceptual capacity. Moreover, by learning from extensive databases of historical performances, these systems can provide stylistic and technical performance suggestions (; ; ). Such suggestions are not intended to offer a single authoritative answer, but rather function as a verifiable and discussable second opinion that provides postgraduate students with an external frame of refere