Moderate evidence
More
University students said AI practice partners built their communication confidence
Study title: Designing AI-mediated interactive learning environments for communication skills: Insights from two implementations
- Evidence type
- Survey
- Sources
- 1 sourcePeer-reviewed
- Published
- 2 min read
- Truvace record
- TRV-2026-3000cite or check it
Study title: Designing AI-mediated interactive learning environments for communication skills: Insights from two implementations
A mixed-methods study looked at two ways universities used AI to help students practise communication skills: a voice-only conversational AI used across health disciplines and a digital human avatar used in a business unit. It combined survey data with qualitative content.
Read the study abstract
Developing communication skills remains a core challenge in higher education, particularly in providing scalable opportunities for practice and feedback. This study explores how artificial intelligence (AI)-mediated systems function as interactive learning environments for communication skills development. Two implementations are examined: a voice-only conversational AI-mediated system used across health disciplines and a digital human avatar used in a business unit. A mixed-methods design was employed, combinin…

55Gain side
How the two sides scoreClose call
52Problem side
The gain
In a mixed-methods study of two university implementations, a voice-only AI used across health disciplines and a digital human avatar in a business unit, students reported increased confidence and perceived skill development.
The problem
In the same two university implementations, students and staff found conversational inaccuracies, limited realism, technical reliability problems and inconsistent feedback from the AI systems.
A fixed formula sets both scores from the evidence and its sources. Reader votes don’t move them.
What the scores mean
G is the gain side, out of 100. It goes up when the gain is specific, when it measurably helped people and when the sources behind it are strong.
P is the problem side, out of 100. It goes up when the problem is specific, when it measurably hurt people and when the sources behind it are strong.
In brief
Students reported more confidence and perceived skill development. They pointed to three benefits: repeated low-stakes practice, immediate feedback and flexible engagement. These are self-reported perceptions, not measured skill gains.
The systems also had weaknesses, including conversational inaccuracies, limited realism, technical reliability problems and inconsistent feedback. The authors found that results depended strongly on prompt design, scenario construction and how well the task matched learner level.
Main points
- Two implementations were examined: a voice-only conversational AI across health disciplines and a digital human avatar in a business unit.
- The study combined quantitative survey data with qualitative content.
- Three benefits emerged: repeated low-stakes practice, immediate feedback and flexible engagement.
- Students reported increased confidence and perceived skill development.
- Reported limitations were conversational inaccuracies, realism, technical reliability and feedback consistency.
- Effectiveness depended strongly on prompt design, scenario construction and alignment with learner level.
The full story
The study addresses a core challenge in higher education: giving students scalable chances to practise communication and receive feedback. It tested AI-mediated systems in two settings, one using voice only across health disciplines and one using a digital human avatar in a business unit.
Data came from surveys and qualitative content. Students described repeated low-stakes practice, immediate feedback and flexible engagement as the main benefits, and reported increased confidence and perceived skill development.
The authors argue AI tools should be treated as designed interactive learning environments needing careful pedagogical integration, not as autonomous technologies. They discuss implications for AI-supported learning activities, particularly oral communication and authentic assessment.
What this doesn’t fix
Reported benefits are students' perceptions of confidence and skill rather than measured skill gains, and the systems had problems with accuracy, realism, reliability and feedback consistency.
Sources
1 on file- ASCILITE Publicationsdoi:10.65106/apubs.2026.3184
Talk it over
Students said repeated low-stakes practice and immediate feedback built their confidence, yet they also met conversational inaccuracies and inconsistent feedback. How do you weigh those benefits against those problems?
University students said AI practice partners built their communication confidencehttps://truvace.com/item/designing-ai-mediated-interactive-learning-environments-for-communication-skills
Students said repeated low-stakes practice and immediate feedback built their confidence, yet they also met conversational inaccuracies and inconsistent feedback. How do you weigh those benefits against those problems?
Answer in the threadThe study combined survey data with qualitative content from a voice-only AI in health disciplines and a digital human avatar in a business unit. What can students' own reports of confidence tell us, and where do they fall short of showing skill?
Answer in the threadIf you were a student practising with the digital human avatar and it gave feedback that changed from one attempt to the next, what would you do with it?
Answer in the thread
If you have more time
The authors found results depended on prompt design, scenario construction and fit with learner level. Who at a university should make those design choices, and how would you check them?
Answer in the threadWhat kind of measurement of oral communication, beyond student surveys, would you want before deciding whether the voice-only AI or the avatar builds skill?
Answer in the thread
ace


What readers say