Skip to main content

GenAI avatars to promote negotiation and AI skills among student presentations

PDF

Abstract

The rapid proliferation of generative artificial intelligence (GenAI) has radically disrupted higher education assessment paradigms (Chung, 2024). Institutional responses to maintaining assessment integrity have primarily focused on defensive mechanisms, such as mitigating risks surrounding written assessments, refining text-matching software, and enforcing restrictive academic integrity policies (Lodge, 2024). However, this reactive posture often overlooks the transformative potential of GenAI as a student’s and lecturer’s long-term coach in authentic assessment design (Atif, 2024). This research-in-progress pivots away from deficit-based AI detection, proposing instead the integration of VirtualSpeech GenAI avatars to converse with students during a recorded viva, generating verifiable, real-time student output. By replacing static written tasks with dynamic oral assessments, this relatively new form of creative expression maintains authentic student contributions to improve AI literacy, strengthen authorship verification, eliminate academic outsourcing, and simultaneously cultivate sophisticated negotiation skills (Perkins, 2026).

This study represents an exemplary innovation of practice by reframing AI not as an academic threat, but as a dual-mode learning and assessment avatar. Rather than engaging with AI purely for text generation, students interact with GenAI avatars (such as VirtualSpeech) simulating realistic corporate stakeholders. This work is strongly anchored in the Technological Pedagogical Content Knowledge (TPACK) framework, which guides the deliberate alignment of GenAI technology, modern pedagogy, and discipline-specific knowledge (Mishra & Koehler, 2006). This framework assists educators in leveraging immersive technology to redefine pedagogical delivery, facilitating AI literacy and negotiation skills. This research addresses a critical gap in current Technology Enhanced Learning (TEL) literature by investigating the following central question: How are students’ AI literacy and negotiation skills developed when conversing with GenAI avatars as part of a viva presentation?

Adopting a design-based research methodology, this project emphasizes the iterative refinement of an educational intervention within an authentic setting (Hoadley & Campos, 2022; Scott et al., 2020). The study is currently in its first implementation cycle within a major tertiary institution, featuring a sample size of 150 second-year Project Management students. Participants are required to defend a project proposal during a live, assessable viva presentation conducted with a customized VirtualSpeech avatar trained to act as a client. This oral format diminishes the appeal of cognitive offloading because students must defend their reasoning under unseen, real-time questioning (Sotiriadou et al., 2020; Tan, 2026). Data collection relies on a mixed-methods approach, utilizing pre- and post-intervention surveys alongside qualitative analysis of student reflections. Preliminary quantitative results indicate a statistically significant increase in both self-reported IT literacy, validating the avatar’s efficacy as a formative and summative tool.

This research significantly advances TEL research and practice by providing a protocol for proactive AI integration. While piloted within a project management curriculum, broad applicability across tertiary contexts exists (Cela-Ranilla, 2025). Any discipline requiring high-stakes communication, critical thinking, and stakeholder management—such as business, law, or education—can adapt this viva model. Ultimately, this work demonstrates how universities can safeguard academic integrity not by restricting technology, but leveraging it to create robust, future-focused verbal assessments, requiring students to perform and defend understanding live.

Keywords

Generative AI, viva, Technological Pedagogical Content Knowledge Framework (TPACK).