AI benefits, challenges, and pedagogical implications on speaking task completion: A multiple case study of two vocational programs in Indonesia

This study investigates spoken English performance in vocational education contexts. Conducted at Politeknik LPP Yogyakarta, a plantation-focused vocational institution, the research adopts a multiple case study design with an embedded quantitative component involving two distinct student groups: 30 Accountancy students and 28 Plantation Crop Farming students. All data were collected with voluntary, informed participation and reported using anonymized participant codes. Findings show that AI use supported time efficiency, task completion, material comprehension, and communicative confidence across both groups. At the same time, the participants showed overreliance on AI-generated language, shallow comprehension of AI-generated content, limited preparation for adapting and internalizing AI-assisted material, and occasional mismatches between AI output and task requirements, leading to the AKT cohort's AI-assisted mean (72.93) falling below its own prior non-AI mean (79.67) as its paradox to the ‘material comprehension’ as AI benefit aforementioned. These findings suggest that AI functions best as a supportive tool rather than a substitute for learner preparation and instructor guidance, particularly in technical vocational fields where accuracy of field-specific vocabulary matters. For the pedagogical implication more broadly, the study highlights the need for structured guidance on AI use alongside speaking skill development. Educators are encouraged to pair AI tools with explicit structured discourse-building and motivation-building strategies. Future research should examine a broader range of AI tools, larger samples, and longer-term effects on spoken language development.