PaperlayerView exact revision on arXiv

Source-linked AI summary

A Comprehensive AI Policy Education Framework for University Teaching and Learning

Cecilia Ka Yuk Chan

arXiv:2305.00280v1cs.CYcs.AI

TL;DR

Universities face concerns about generative AI misuse and limited policy guidance for its educational integration. Using quantitative and qualitative data from Hong Kong university stakeholders, the study proposes a three-dimensional AI Ecological Education Policy Framework.

  • Problem

    Limited policy guidance and concerns about generative AI misuse, academic integrity, and ethical implications motivate a comprehensive higher-education policy framework.

  • Method

    The study combines descriptive survey analysis with qualitative and quantitative data from students, teachers, and staff across Hong Kong universities.

  • Results

    The study proposes an AI Ecological Education Policy Framework spanning Pedagogical, Governance, and Operational dimensions, reflecting stakeholders’ concerns and support for institutional planning.

  • Takeaways & Limitations

    The framework assigns stakeholder responsibilities for responsible AI integration, with students participating actively in policy drafting and implementation.

  • Takeaways & Limitations

    The study calls for further research because AI’s educational benefits, risks, personalization, and ethical dilemmas remain insufficiently understood.

Abstract

from arXiv · show

This study aims to develop an AI education policy for higher education by examining the perceptions and implications of text generative AI technologies. Data was collected from 457 students and 180 teachers and staff across various disciplines in Hong Kong universities, using both quantitative and qualitative research methods. Based on the findings, the study proposes an AI Ecological Education Policy Framework to address the multifaceted implications of AI integration in university teaching and learning. This framework is organized into three dimensions: Pedagogical, Governance, and Operational. The Pedagogical dimension concentrates on using AI to improve teaching and learning outcomes, while the Governance dimension tackles issues related to privacy, security, and accountability. The Operational dimension addresses matters concerning infrastructure and training. The framework fosters a nuanced understanding of the implications of AI integration in academic settings, ensuring that stakeholders are aware of their responsibilities and can take appropriate actions accordingly.

1. Introduction

The introduction presents generative AI as both a threat to academic integrity and student learning and a potential means of improving personalized educational support. It argues that universities need an evidence-informed AI education policy for teaching and learning that addresses implementation gaps and incorporates stakeholder input.

  • Risks and concerns: Generative AI raises concerns about cheating, plagiarism, academic misconduct, and declining writing and critical-thinking skills.These concerns have prompted calls for stricter penalties and bans or policy reviews at some universities.
  • Potential benefits: Generative AI may also enhance education through personalized, real-time feedback and adaptive support that helps students identify weaknesses and improve their skills.The introduction therefore frames AI integration as involving both risks and potential learning benefits.
  • Rationale for policy: Universities need AI education policies because AI is becoming prevalent across economic sectors and increasingly embedded in education, assessment, and society.Such policies can prepare students and teachers to use AI professionally, understand its principles, and navigate related ethical, social, and economic issues.
  • Policy limitations: Existing AI policies often emphasize broad ethical and legal principles while remaining generic, implicit, or insufficiently grounded in concrete implementation evidence.This may limit effective governance and overlook AI in education’s transformative potential and ethical implications.
  • Study purpose: The study uses UNESCO recommendations as a starting point for developing an AI education policy framework for university teaching and learning.The framework will be adapted using recommendations, ideas, concerns, and gap-identification from various stakeholders.

2. Methodology

The study used an online survey to examine Hong Kong education stakeholders’ usage and perceptions of generative AI in teaching and learning. It combined quantitative and qualitative analyses to identify requirements and strategies for university AI policy.

  • Survey design: The online questionnaire examined generative AI usage, higher-education integration, and potential risks among Hong Kong students, teachers, and staff.It included both closed-ended and open-ended questions, including items about ChatGPT use and envisioned teaching and learning applications.
  • Participants and sampling: 457 students and 180 teachers and staff across various disciplines in Hong Kong completed the survey.Respondents were selected through convenience sampling based on availability and willingness to participate, and provided informed consent.
  • Data analysis: Quantitative survey data were analysed descriptively to summarize the main characteristics of participants and their generative AI usage and perceptions.Descriptive analysis was used for survey responses from students and teachers in Hong Kong.
  • Data analysis: Open-ended responses on apprehensions and university strategic-plan recommendations were analysed thematically to identify patterns and themes across disciplines.Combining quantitative and qualitative data supported a more holistic understanding and helped pinpoint requirements, recommendations, and strategies for AI policy.

3. Results

Survey responses from 457 students and 180 teachers and staff showed openness to generative AI in higher education alongside concerns about ethical use, over-reliance, fairness, and teachers’ ability to detect AI use. The findings support comprehensive institutional policies combining training, risk management, and AI as a supplement to human interaction.

  • Quantitative results: 457 students and 180 teachers and staff participated in a Hong Kong university survey examining requirements, guidelines, and strategies for AI teaching-and-learning policies.The survey covered different disciplines and explored perceptions of generative AI technologies such as ChatGPT.
  • Quantitative results: Students reported low generative-AI experience (mean=2.28, SD=1.18), as did teachers (mean=2.02, SD=1.1), while both groups believed AI could positively affect higher education.Students’ positive-impact belief was mean=4, SD=0.891; teachers’ was mean=3.87, SD=1.32.
  • Quantitative results: Students and teachers were open to future AI integration, but both were concerned that other students could use AI to gain an assignment advantage.Openness was students: mean=3.93, SD=1.09; teachers: mean=3.92, SD=1.31. Concern was students: mean=3.67, SD=1.22; teachers: mean=3.93, SD=1.12.
  • Quantitative results: Both groups valued learning generative AI for careers but doubted teachers’ ability to accurately identify students’ AI use in assignments.Career importance was students: mean=4.07, SD=0.998; teachers: mean=4.1, SD=1.08. Detection confidence was students: mean=3.02, SD=1.56; teachers: mean=2.72, SD=1.62.
  • Policy implications: Respondents recognized benefits including guidance, personalized feedback, digital competence, academic performance, and anonymous support, while identifying risks including over-reliance, reduced social interaction, and weaker generic skills.The results motivate comprehensive policy covering training, ethical use, risk management, and AI as a supplementary rather than replacement tool for human interaction.

(1) Understanding, Identifying and Preventing Academic Misconduct and Ethical

Universities should establish clear rules and educational strategies to prevent generative-AI-related academic misconduct and strengthen students’ understanding of ethical boundaries. AI governance should also require transparency, accountability, privacy, and security protections, including disclosure of algorithms, limitations, and data practices.

  • Understanding, Identifying and Preventing Academic Misconduct and Ethical: Universities should create clear policies and strategies for detecting and preventing misuse of generative AI in prohibited student tasks.Teachers emphasized university-wide rules for testing suspected misuse and resources that inform students about applicable requirements.
  • Understanding, Identifying and Preventing Academic Misconduct and Ethical: Students and staff should be familiarized with ethical dilemmas, including boundaries between plagiarism and inspiration and appropriate situations for seeking AI assistance.The passages call for stronger education on academic and research ethics and clear ethical and legal responsibilities.
  • Addressing Governance of AI: Data Privacy, Transparency, Accountability and: Universities should disclose how generative AI is implemented, including algorithms, functions, potential biases, and limitations, while remaining receptive to feedback and criticism.Transparency about implementation is presented as a means of fostering trust and confidence among students and staff.
  • Addressing Governance of AI: Data Privacy, Transparency, Accountability and: Institutions should keep data used by generative AI private and secure, de-identify training and testing data, and prevent unauthorized access or use.The passage specifically links these safeguards to concerns arising from AI technologies’ reliance on vast amounts of data.
  • Addressing Governance of AI: Data Privacy, Transparency, Accountability and: Institutions should address discrimination, bias, stereotypes, privacy, and security while clarifying accountability for decisions and actions involving complex AI technologies.The complexity of AI can make accountability difficult for both organizations and individuals.

(3) Monitoring and Evaluating AI Implementation

Successful AI integration in university teaching and learning requires continuous monitoring and evaluation. Teachers recommend longitudinal experiments and regular assessments to understand AI’s effects, identify improvements, and support effective and ethical use.

  • Monitoring and Evaluating AI Implementation: Continuous monitoring and evaluation are necessary to ensure successful AI integration in university teaching and learning.Regular evaluation helps determine whether implementation is achieving its intended purposes.
  • Monitoring and Evaluating AI Implementation: Teachers recommend longitudinal experiments across different areas to examine AI’s effects on students’ learning processes and outcomes.Studying implementation over time and across contexts can clarify how AI influences learning.
  • Monitoring and Evaluating AI Implementation: Regular assessments of AI’s impact on teaching practices and student performance can identify areas for improvement and support effective, ethical use.The assessments should consider both instructional practice and student performance.

(4) Ensuring Equity in Access to AI Technologies

The section emphasizes that equitable access to AI technologies is essential for inclusive and fair university education. Universities should provide AI resources and support broadly while preventing discriminatory use and unequal access in competitive contexts.

  • Equitable access: Universities should provide AI resources and support to all students and staff, regardless of background or technology access.This may include procuring AI tools, including AI detectors, for the entire university community.
  • Fairness and non-discrimination: Equal access to AI technologies should be a top priority for fairness in education, particularly when technology use involves competition.A teacher specifically identified equal access to ChatGPT for all involved parties as an example.
  • Fairness and non-discrimination: Universities should ensure that AI technology is not used to discriminate against individuals.A student identified non-discrimination as an ethical dilemma associated with AI use.

(5) Attributing AI technologies

The section presents attribution as a central element of AI policy in university teaching and learning. It recommends clear disclosure of AI-assisted contributions and guidelines for fair attribution of generative AI use.

  • Attribution: Universities should require students to clearly identify which parts of their academic work were assisted by AI.This approach is likened to established academic referencing and citation practices.
  • Attribution: Clear attribution of AI-generated content can help universities promote academic integrity.
  • Attribution: Fair attribution guidelines should address AI’s ethics of use, affordances, effective use, output evaluation, and role in academic and professional workflows.

(7) Rethinking Assessments and Examinations … (10) Developing Student Holistic Competencies/Generic Skills

The paper calls for reassessing examinations, adopting generative AI cautiously, and preparing students for AI-enabled workplaces while strengthening critical thinking, ethics, and broader holistic competencies.

  • (7) Rethinking Assessments and Examinations: Assessments should use AI to enhance learning outcomes rather than merely produce outputs, requiring universities to rethink examinations and assessment design.Teachers recommend activities that help students discover AI’s limitations and question its use as a means of cheating.
  • (8) Encouraging a Balanced Approach to AI Adoption: AI adoption should balance potential gains in efficiency and productivity with critical thinking, ethical considerations, flexibility, and safeguards against over-reliance.Teachers advocate incorporating AI into new assignments and assessments while ensuring it assists rather than replaces students.
  • (9) Preparing Students for the AI-Driven Workplace: Universities should prepare students for AI-driven workplaces by teaching responsible, ethical, and effective AI use alongside curricula reflecting AI’s growing prominence across industries.Students should learn to integrate AI into workflows, evaluate tools, and understand their professional role.
  • (9) Preparing Students for the AI-Driven Workplace: Students should become familiar with AI tools used in education and work, recognize ethical issues, and learn to appropriate these tools for study and professional settings.One student proposed making AI tools a common part of education, like PowerPoint and Excel.
  • (9) Preparing Students for the AI-Driven Workplace: Universities should support constructive student use of AI in learning, career planning, and personal development as workplaces increasingly adopt these technologies.Students should receive plans and assistance for navigating this shift.
  • (10) Developing Student Holistic Competencies/Generic Skills: Teaching should develop critical thinking, digital and information literacy, and professional ethics so students can assess AI-generated content’s reliability, biases, accuracy, and relevance.Students also need support maintaining motivation for deep thinking, diversifying perspectives, and expanding horizons.
  • (10) Developing Student Holistic Competencies/Generic Skills: AI integration may hinder teamwork, leadership, empathy, creativity, and other holistic competencies, so universities must preserve opportunities to develop adaptability and resilience.Broader educational aims also include character, rhetoric, analytical skills, public speaking, memorisation, and embodied skills.

4. Discussion

The discussion links shared concerns and expectations about generative AI to the need for equitable access, AI literacy, revised assessment, and balanced adoption. It also argues that a stakeholder-informed framework addresses university-specific gaps left by broader UNESCO recommendations.

  • Implications for AI integration: Students and teachers reported concern that generative AI could be misused in assignments, with means of 3.67 and 3.93, respectively, underscoring the need for academic-misconduct guidelines.The concern concerns potential misuse of technologies such as ChatGPT in assignments.
  • Implications for AI integration: Students and teachers agreed that students must become proficient with generative AI for their careers, with means of 4.07 and 4.1, respectively, supporting AI literacy and training.The discussion also connects possible assignment advantages from AI use to ensuring equal access for all students.
  • Implications for AI integration: Because students and teachers were unsure whether teachers could accurately identify partial AI use in assignments, with means of 3.02 and 2.72, the discussion suggests rethinking assessment methods.Both groups also believed AI could provide unique insights, perspectives, and personalized feedback.
  • Implications for AI integration: Students and teachers did not believe AI would replace teachers, with means of 2.14 and 2.26, supporting adoption of AI as a complement rather than a substitute for teaching.The discussion also raises concern that generative AI could hinder transferable skills such as teamwork, problem-solving, and leadership.
  • Policy framework development: The ten stakeholder-informed areas differ from UNESCO recommendations because they are more relevant to university teaching and learning and address current GPT-related opportunities and threats.UNESCO guidance was developed before GPT 3.5 and 4 and is broader and more general, whereas the study’s assessment area is more specifically targeted to student evaluation.
  • AI Ecological Education Policy Framework: The ten key areas were organized into Pedagogical, Ethical, and Operational dimensions, each led by a responsible party, to translate policy recommendations into action plans.The framework is intended to support a nuanced understanding of AI’s multifaceted implications in university settings and broader adoption impacts.

Pedagogical Dimension (Teachers) · Governance Dimension (Senior Management)

The Pedagogical Dimension guides teachers in adapting teaching and learning to AI integration, while the Governance Dimension addresses ethical, accountability, privacy, security, and equity considerations for AI use in education.

  • Pedagogical Dimension (Teachers): Teachers should rethink assessments and examinations as AI becomes integrated into teaching and learning.
  • Pedagogical Dimension (Teachers): The pedagogical approach should develop students’ holistic competencies and generic skills.
  • Pedagogical Dimension (Teachers): Teachers should prepare students for the AI-driven workplace while encouraging a balanced approach to AI adoption.
  • Governance Dimension (Senior Management): Senior management should understand, identify, and prevent academic misconduct and ethical dilemmas linked to AI use.
  • Governance Dimension (Senior Management): AI governance should address data privacy, transparency, accountability, and security.
  • Governance Dimension (Senior Management): Governance should include attribution of AI technologies and ensure equity in access to them.

Operational Dimension (Teaching and Learning and IT staff)

The Operational dimension focuses on practical AI implementation through monitoring, evaluation, training, support, and continuous adaptation. Teaching and Learning and IT staff oversee these efforts to promote effective, equitable, and minimally disruptive integration.

  • Operational Dimension: The Operational dimension requires monitoring and evaluating AI implementation while providing AI-literacy training and support for teachers, staff, and students.These activities address practical implementation across university settings.
  • Operational Dimension: Ongoing monitoring, evaluation, training, resources, and support promote effective and equitable implementation of AI technologies across stakeholders.The framework also seeks equal access and an inclusive learning environment.
  • Operational Dimension: Continuous improvement and adaptation enable universities to refine AI-integration strategies in response to new insights and changing needs.Operational planning is therefore treated as an ongoing process rather than a one-time implementation.
  • Operational Dimension: Teaching and Learning and IT staff manage and maintain educational AI technologies, provide training and support, and address technical issues to minimize disruptions.Their responsibilities help integrate AI technologies seamlessly into the educational environment.
  • Operational Dimension: Successful policy implementation requires collaboration and communication among universities, teachers, students, staff, and external agents.All stakeholder groups should participate actively in developing and executing AI-related initiatives.

5. Conclusions

The study proposes an AI Ecological Education Policy Framework for university teaching and learning, addressing the diverse implications of text-generating AI integration. It emphasizes responsible, ethical implementation while acknowledging limitations and the need for further research into AI’s benefits, risks, technologies, methods, and capabilities.

  • Limitations and future research: The study acknowledges limitations including a relatively small sample size that may not be representative.The conclusion presents these limitations alongside findings drawn from quantitative and qualitative data involving various stakeholders.
  • Framework contribution: The study proposes an AI Ecological Education Policy Framework addressing the diverse implications of AI integration in university settings.The framework responds to concerns about text-generating AI in academic environments, including cheating and plagiarism.
  • Framework contribution: The framework comprises Pedagogical, Governance, and Operational dimensions, each led by a responsible party.This structure promotes comprehensive understanding of AI integration and clarifies stakeholder responsibilities.
  • Implementation implications: Institutions can use the framework to align actions with policy and support responsible and ethical AI usage while maximizing potential benefits.The conclusion links framework adoption with policy-aligned institutional action.
  • Limitations and future research: Further research is needed to clarify AI’s potential advantages and risks and to explore its untapped potential to transform learning.The conclusion states that advocating AI implementation alone is insufficient.
  • Limitations and future research: Stakeholders must evaluate which AI technologies to employ, determine suitable usage methods, and understand their true capabilities.These decisions are identified as necessary for informed AI implementation in education.
Loading 2305.00280v1…