Validating A Technology Acceptance Model of Generative AI in Undergraduate Music Education

Abstract

Generative artificial intelligence (AI) tools such as ChatGPT are rapidly entering higher education, yet their acceptance in highly embodied and practice‑oriented domains like music education remains underexplored. This study examines relationships among perceived usefulness (PU), perceived ease of use (PEOU), attitude toward use (ATT), behavioral intention (BI), and actual use (AU). A cross‑sectional survey of 218 music education students was conducted, supported by focus group discussions (FGDs) and classroom observations. A 28‑item TAM‑based instrument was adapted to Generative AI in music learning and underwent expert review, pilot testing, and full validation. Measurement results indicated satisfactory reliability and validity (Cronbach’s α = 0.80–0.89; composite reliability = 0.84–0.91; average variance extracted = 0.56–0.66; HTMT < 0.85), with good model fit (χ²/df = 2.11, CFI = 0.953, TLI = 0.943, RMSEA = 0.072, SRMR = 0.049). Descriptive results suggested generally positive acceptance (means on a 1–5 scale: PU = 4.07, PEOU = 3.94, ATT = 4.02, BI = 3.88, AU = 3.41). Students predominantly used AI to summarize theory, brainstorm composition ideas, generate practice drills, and simplify technical terms, while performative and practical uses remained limited. The findings corroborate TAM’s applicability to Generative AI in music education and resonate with evidence from language learning and teacher education contexts. We propose institutional strategies to support responsible adoption: concise usage guidance, assessment‑integrated AI literacy, and transparent ethical policies addressing originality and authorship. The study offers a domain‑specific, validated instrument and empirically grounded recommendations for integrating Generative AI as a cognitive and creative adjunct—rather than a replacement—for embodied musical learning.