AI evidence extraction

Supervision Skills of Supervisors in AI-enhanced Environments: Perspectives on Postgraduate Supervision

Thembi Busisiwe Nkosi · Open Books and Proceedings · 2026

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

9/10

Relevance

1/4

Quality (LMQS)

Evidence

Citations

20.68

FWCI

Top 10%

Impact

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.38140/obp4-2026-06

Methodology & findings

Study design

Exploratory phenomenology using qualitative research methods.

Sample

N = 10, 6 groups

Primary method

Qualitative thematic analysis. Transcripts underwent multiple examinations to detect initial patterns. Open coding was used to assign initial labels to data sections, which were then organised based on conceptual meanings and compared to the full dataset. No quantitative statistical methods reported.

Main result

The study identified three major themes of supervisory skills: (1) Critical Evaluation Skills - supervisors must "read between the lines to identify when AI is using their students" and detect improper AI use through in-text referencing and source acknowledgment; (2) Pedagogical Mentorship Skills - supervisors need "digital literacy, critical thinking, ethical awareness, and adaptability" and must "train them how to use it so that when using it they can use it appropriately"; (3) Adaptive Communication Skills - supervisors must provide "clear guidance to students on the appropriate use of AI tools in academic work" rather than remaining silent. The convergence of all skills around academic integrity suggests that "effective supervision requires a proactive and comprehensive framing of AI issues, from detection to training."

Reports effect sizes.

Research paradigm

Interpretive/constructivist paradigm

Author conclusions

"The study concluded that supervisors have been positioned as crucial gatekeepers who must navigate the interplay between technology and human intuition, ensuring that AI support does not override student autonomy. This places them in a pivotal role as protectors of the student experience, with mentorship serving as a critical safeguard against over-reliance on AI." Furthermore, "the overarching inference from the entire study suggests a shift in the landscape of postgraduate supervision, where AI not only changes the tools available but also potentially alters the epistemology of supervision. This necessitates an ongoing effort in supervisor upskilling to model the behaviours and skills necessary for critical digital citizenship."

Risk of bias

Selection bias: Purposive sampling of only 10 supervisors may not be representative of broader supervisor populations; Geographic limitation: Study restricted to three South African universities, limiting generalizability; Self-report bias: Data based on supervisor interviews; no triangulation with student perspectives mentioned; Small sample size reduces transferability of findings; No mention of inter-rater reliability or data validation procedures in thematic analysis; Selection bias: Purposive sampling of supervisors may not represent all supervisory perspectives; Small sample size (n=10) limits generalizability; Geographic limitation to South African institutions; Potential self-report bias in semi-structured interviews; Researcher interpretive bias in thematic analysis; Selection bias: Purposive sampling of only 10 supervisors from 3 South African universities limits generalizability; Potential interviewer bias in semi-structured interviews; No discussion of researcher reflexivity or positionality; No mention of inter-rater reliability for thematic analysis; Geographic limitation to South African context only; No triangulation of data sources mentioned

Open questions raised

Need for specialised training programmes to develop supervisory skills for AI technology management
Lack of existing frameworks for preparing supervisors who lack direct experience with tasks before AI integration
Limited understanding of how supervisors can preserve student independence and critical thinking while integrating AI tools
Insufficient guidance on establishing explicit guidelines for responsible AI use in academic research
Need for supervisor training on AI literacy and ethical considerations in mentorship approaches
Research should focus on creating scalable supervisor training frameworks that meet unique demands of AI in postgraduate supervision across different academic disciplines and institutional environments

Extracted from: pdfAgreement 68%

Explore related topics

AI Literacy Training Doctoral Training Qualitative