AI evidence extraction

Challenges and Opportunities of Generative AI for Higher Education as Explained by ChatGPT

Rosario Michel‐Villarreal, Eliseo Luis Vilalta-perdomo, David Ernesto Salinas-Navarro, Ricardo Thierry-Aguilera, Flor Silvestre Gerardou · Education Sciences · 2023

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

5/10

Relevance

Evidence

749

Citations

26.55

FWCI

Top 10%

Impact

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.3390/educsci13090856

Methodology & findings

Study design

Thing ethnography applied to ChatGPT using semi-structured interview methodology.

Main result

The study found that ChatGPT presents both significant opportunities and challenges for higher education. According to the interview, "ChatGPT indeed presents several significant benefits and opportunities for higher education" including 24/7 support, personalized learning, supplemental resources, language learning support, instructor support, innovative learning experiences, and research assistance. However, the findings also revealed that "ChatGPT, as an advanced language model, presents certain challenges for higher education" such as academic integrity concerns, quality control issues, limitations in personalized learning, lack of domain expertise, and reduced human interaction. The study emphasizes that "vast challenges need to be addressed first" before full integration into educational settings.

Research paradigm

Interpretivist/Constructivist

Author conclusions

The authors conclude: "The findings of this study highlighted the transformative potential of ChatGPT in education, consistent with previous studies, while also revealing additional insights. It also highlighted significant challenges that must be addressed." They further state: "A key takeaway point is the urgent need for empirical research that delves into best practices and strategies for maximizing the benefits of GenAI, as well as user experiences, to understand students' and academics' perceptions, concerns, and interactions with ChatGPT. Another priority is the development of policies, guidelines, and frameworks for the responsible integration of ChatGPT in higher education." The authors also argue that "ChatGPT should be approached not merely as an object controlled by humans but as a subject actively engaging in daily practices alongside users, influencing communication dynamics, and impacting socio-material networks."

Risk of bias

Selection bias: Study focused only on ChatGPT, not other GenAI chatbots; Replicability concerns: Future versions of ChatGPT may provide different responses; Research bias during analysis: Potential researcher interpretation bias in thematic analysis; Limited longitudinal design: Cross-sectional design, single point in time data collection; findings may not generalize to other GenAI chatbots like Bard or Bing Chat; Prompt-dependent responses: ChatGPT's answers depend heavily on how questions are framed by the interviewer; Lack of human participant validation of ChatGPT's self-described perspectives; Interview conducted with language model that may exhibit self-presentation bias

Limitations

The authors acknowledge several limitations: "The first one relates to the limited scope and selection bias
ethnographic research usually focuses on specific individuals, groups, or communities
In this study, there is a strong focus on ChatGPT as a form of GenAI
However, other GenAI chatbots, such as Bard or Bing Chat, have recently emerged and become accessible to the public
Thus, the findings presented here may not be representative of the views of all existing GenAI chatbots." Additionally, "Another limitation could be the replicability of results
even though the interview can be easily replicated, it is still not clear that ChatGPT will provide the same answers." The authors also note that "ethnographic studies tend to be highly immersive and lengthy, requiring extended periods in the field to gain an in-depth understanding of the research subject

Open questions raised

Need for empirical research on best practices and strategies for maximizing benefits of GenAI
Need to understand user experiences and perceptions of students and academics
Gap in knowledge about transferability of findings to varied contexts (e.g., distance or online learning)
Need to compare views from different GenAI chatbots to find similarities and discrepancies
Need for longitudinal research to test stability of findings across ChatGPT versions
Need for further conversations with ChatGPT and other chatbots to test replicability

Data: The authors state "A full transcript of the interview with ChatGPT is included in this paper. No additional data was created for this study."Extracted from: pdf

Explore related topics

Responsible AI Use Doctoral Training Qualitative