Digital Literacies
November 29, 2025
Preface
Artificial Intelligence (AI) is reshaping higher education by providing new tools for writing,
research, instruction, student support, and institutional management. This transformation
creates both opportunities and obligations. The pedagogical promise of AI includes improved
access to information, personalized learning pathways, scalable formative feedback, and faster
data analysis for research. However, AI’s affordances also bring ethical dilemmas: the capacity
to generate credible but false information (hallucinations), the possibility of hidden bias in
outputs that reproduce societal inequities, privacy and data-protection risks when personal or
research data are uploaded to third-party platforms, and copyright issues when generative
models reuse or reproduce protected creative content.
This mini-module translates the theoretical concepts of Materi #3, Etika Penggunaan AI into practical materials specifically designed for the Universitas Muhammadiyah Purwokerto (UMP) community. It presents a robust ethical framework rooted in the UNESCO and OECD principles, explains relevant Indonesian laws (UU ITE, UU PDP, UU Hak Cipta), provides operational guidelines for students and lecturers, includes a narrative ethical decision flow, offers a visual integrity pyramid, and presents three campus-relevant case studies with detailed ethical, legal, and remedial analyses. The module aims to be directly usable in classrooms, workshops, orientation sessions, and research training programs.
Introduction to AI
Artificial Intelligence, for purposes of this module, denotes computational systems that
perform tasks typically associated with human cognitive functions, including classification,
prediction, language processing, pattern recognition, and generative production of text, images,
audio, or code. Over the last decade, deep learning approaches, particularly transformer-based
architectures and large-scale pretraining, have produced generative AI systems that can write
essays, summarize literature, translate languages, synthesize images, and draft code. The
modern development of artificial intelligence can be traced from Alan Turing’s foundational
1950 inquiry, which introduced the conceptual basis for machine reasoning, to the emergence
of generative AI systems in 2022 that marked a new era of large-scale models capable of
producing human-like text, images, and other modalities.
To make practical pedagogical distinctions, we classify AI into three types. Narrow AI refers
to systems designed to perform a specific task; examples include plagiarism-detection
software, recommender systems, and rule-based tutoring systems. Generative AI, the most
relevant to current classroom challenges, generates novel outputs conditioned on prompts:
examples include large language models (LLMs) for text and diffusion models for images.
Artificial General Intelligence (AGI) remains theoretical and is not central to current
educational policy.
Key technical behaviors that bear on academic use include:
Probabilistic generation: LLMs do not know facts; they predict tokens based on learned
patterns, producing fluent text that can nonetheless be inaccurate.
Hallucination: the tendency to produce plausible-sounding but false statements or
fabricated citations, presenting a critical risk in academic contexts.
Bias reproduction: models mirror and sometimes amplify biases present in their training
data, which can reproduce stereotypes or marginalize groups unless specifically audited
and corrected.
Data leakage and privacy risk: models trained or queried with sensitive data may reveal
or memorize personal information, posing legal and ethical concerns.
Understanding these behaviors is essential. AI can become a powerful educational scaffold
when its limitations are recognized and mitigated through careful human oversight and
verification.
Types of AI (Narrow AI, Generative AI, AGI)
Narrow AI systems are embedded across campus systems (e.g., library recommendation
engines, plagiarism detection tools, automated timetable scheduling). Their logic tends to be
transparent and task-specific. They are typically lower risk if used within intended scopes,
though they still require oversight for fairness and robustness.
Generative AI systems, such as LLMs (e.g., ChatGPT-style models) and image generators (e.g.,
diffusion-based models), are capable of producing rich, human-like outputs. From Materi #3,
examples include automatic summarization tools that condense long texts and conversation
agents that help students restructure arguments. Their high utility for drafting and
brainstorming is counterbalanced by risks: fabricated references, overconfidence, and potential
licensing or copyright concerns when outputs replicate protected content.
AGI remains an aspirational concept indicating systems with general problem-solving
capabilities comparable to humans. Because AGI does not currently exist in deployed academic
tools, policy should focus on narrow and generative AI while monitoring research
developments.
Benefits for Education & Research
AI can enhance learning and research in multiple, evidence-backed ways. First, personalized
learning: adaptive learning platforms can tailor content and pacing to individual students,
providing remediation or acceleration where needed. Second, accessibility: AI-powered
translation and text simplification tools assist non-native speakers and students with differing
learning needs. Third, productivity and scaffolding: AI can generate initial drafts, suggest
structure, propose literature keywords, or create visual aids, thereby lowering entry barriers for
research and enabling students to focus on critical thinking and conceptual refinement. Fourth,
analytic acceleration: AI can process large datasets (e.g., text corpora, survey responses)
to identify patterns, which assists researchers in hypothesis generation and exploratory
analysis.
However, realizing these benefits requires pedagogical redesign: assignments and assessments
must be structured to reward process over product, requiring documentation of steps, iterations,
and verification. In research, AI must be integrated into ethical research design, with careful
data governance and methodological transparency.
Risks (Bias, Deepfake, Plagiarism, Privacy Violation)
The principal risks of AI in academia include the following:
Hallucination and misinformation. Generative AI may fabricate facts, dates, or
references with fluent prose. When students or researchers use such outputs uncritically,
they risk spreading falsehoods and compromising academic standards.
Plagiarism and academic integrity subversion. AI-assisted rewriting or full-generation
of essays can obscure authorship. Traditional plagiarism detectors may fail to detect
paraphrased or synthesized outputs, demanding updated academic integrity policies and
process-based assessment.
Deepfakes and image misuse. Generative image and audio tools can create convincing
synthetic media. When used to manipulate student likenesses or produce deceptive
content, these tools can inflict reputational and psychological harm and may contravene
national laws.
Algorithmic bias and unfairness. If AI is used in high-stakes contexts (e.g., formative
grading, recommendations for scholarships), biased outputs can systematically
disadvantage particular groups. Institutions must audit models and monitor outcomes
for disparate impacts.
Privacy and data protection risks. Uploading student data, interview transcripts, or
identifiable research data to third-party AI platforms may result in unauthorized
retention or reuse. Such acts can violate the UU PDP and ethical norms for research
subject protection.
Each risk requires concrete institutional mitigation: pedagogical redesign, mandatory
verification steps, consent protocols, approved tool registries, contractual safeguards with
vendors, and ongoing auditing.
Legal Responsibilities (UU ITE, UU PDP, UU Hak Cipta)
UU ITE (Information and Electronic Transactions). This law addresses the legal
responsibilities around electronic communications, including prohibitions related to
hoaxes, defamation, and the unlawful manipulation of electronic content. In practice,
creating or distributing AI-generated defamation, hoaxes, or deepfakes that harm
individuals or the university can fall under UU ITE. Universities must educate staff and
students about legal exposure and implement content review protocols for public
communications.
UU PDP (Personal Data Protection). This law formalizes individual rights and data
controller responsibilities: consent, purpose limitation, data minimization, security,
retention policies, and procedures for breaches. For campus practice, researchers and
instructors must secure explicit consent for collecting and processing personal data,
especially when third-party AI services are involved. Contracts with AI vendors must
include data processing agreements that prohibit retention or reuse of academic data for
model training unless explicitly permitted.
UU Hak Cipta (Copyright Law). Copyright law protects creative works. AI outputs
that reproduce copyrighted text, images, or derivative content may raise infringement
concerns. Additionally, training datasets and model outputs can implicate copyright if
provenance is unclear. Universities must ensure licensed usage of copyrighted works
for training or teaching and instruct students not to prompt AI in ways that generate
copyrighted content for reuse without appropriate licenses.
Combining these legal frameworks, institutions should adopt procurement policies, consent
forms, data protection impact assessments, and researcher training to ensure compliance and
reduce legal risk.
Ethical Principles (UNESCO, OECD), Applied to Campus Practice
Both UNESCO and OECD frameworks emphasize human-centric and trustworthy AI.
Applying these to the UMP campus involves several operational principles:
Transparency and explainability. Students and lecturers should document AI use, and
institutions should publish which AI tools are approved and why. When AI informs
assessment or administrat ive decisions, explanations must be provided and appeal
processes put in place.
Fairness and non-discrimination. AI systems used in admissions, assessment, or student
support must be audited for disparate impact. Institutions should avoid automating
decisions that have unequal effects across demographic groups without human review.
Human oversight and accountability. AI should augment, not replace, human judgment.
Faculty must remain responsible for grading and evaluative decisions, and researchers
must take responsibility for data processing outcomes.
Data governance and privacy. Institutions must institute strict data policies consistent
with UU PDP, including anonymization, secure storage, and vendor contracting.
Sustainability and societal benefit. AI adoption should serve educational goals and the
public good, not merely operational efficiency.
These principles are translated into policies, training modules, and practical checklists to ensure
alignment with global best practice.
Practical Do’s & Don’ts for UMP Students & Lecturers
Do’s (detailed):
Disclose AI use. For any assessed work or research, include a short disclosure
statement specifying the tool (name and version), the function it performed (idea
generation, editing, summarizing), and the extent of human revision.
Verify and cite primary sources. Treat AI outputs as preliminary. Cross-check
facts, locate and read primary sources, and cite original literature, not the AI tool.
Document process. Retain prompt histories, drafts, and revision logs
to demonstrate learning and intellectual contribution.
Use institutional tools. Prefer university-approved platforms that have vetted data
protection arrangements.
Obtain informed consent. For research involving human subjects, include explicit
consent for any data processing involving AI, document consent forms, and describe
AI use in ethics submissions.
Design assessments for learning. Employ staged submissions, oral components, and
reflective elements to evaluate process and understanding.
Don’ts (detailed):
Do not submit entirely AI-generated work as original. Passing off machine-generated
content as one’s own is academic misconduct.
Do not accept AI-generated citations at face value. AI may invent sources.
Do not upload identifiable or sensitive data to third-party tools without approval. This
includes student grades, interview transcripts, or image files.
Do not create or distribute deepfakes or manipulated content of peers or staff. This risks
legal action and severe disciplinary measures.
Do not rely on AI for final evaluative judgments without human review. Automated
systems should be advisory and transparent.
These rules should be embedded in syllabus statements, honor codes, research ethics forms,
and IT acceptable-use policies.
Ethical AI Use Flowchart
The following narrative flow acts as a decision checklist for individuals considering AI usage
in academic tasks:
- Define the Task and Learning Objective. Start by explicitly defining the academic
purpose. Is the task formative or summative? Is the learning objective process-oriented
(developing reasoning) or product-oriented (producing polished text)? - Check Institutional Policy. Consult course and university policies for explicit
allowances or prohibitions regarding AI for the particular task. If prohibited, do not use
AI. - Specify AI Role. If permitted, decide whether the AI will serve as a brainstorming tool,
language editor, literature summarizer, code aid, or other defined role. - Assess Data Sensitivity. Determine whether the task requires handling personal,
identifiable, or sensitive data. If yes, stop and obtain approvals and legal/ethical
clearance. - Identify Potential Risks. List risks such as hallucinations, bias, privacy breaches, and
copyright infringement. Consider mitigation strategies (independent verification,
anonymization, alternate datasets, human-in-the-loop review). - Select Approved Tools and Safeguards. Use institutionally approved platforms
whenever possible, or obtain a data processing agreement with any third-party
provider. - Generate AI Output with Constraints. Use carefully framed prompts, avoid exposing
private data, and limit the AI’s role to agreed tasks. - Human Review and Verification. Review every claim, check references, perform fact
checking, and adapt outputs to align with disciplinary norms. - Document and Disclose. Archive prompt history and revision logs, and include
disclosure in submissions or publications. - If Unacceptable Risk Remains, Abort. If mitigation does not sufficiently reduce risk,
revert to human-only methods.
This flow ensures auditable decisions and reduces the likelihood of ethical or legal breaches.
Visual Framework: AI Academic Integrity Pyramid
The pyramid is a layered model representing sequential responsibilities:
Base: AI Literacy & Consent. Foundational knowledge includes understanding model
types, common failure modes, data protection basics, and consent principles. Training
modules should be provided to develop this literacy among all stakeholders.
Second Layer: Responsible Use & Verification. Users must not accept outputs at face
value. Verification routines (source triangulation, database searches, peer consultation)
should be routine. Responsible use also encompasses prompt design that avoids
invoking copyrighted or sensitive content.
Third Layer: Transparency & Attribution. Disclosure mechanisms should be
standardized. Instructors should require students to declare AI use; researchers should
include AI methods in methods sections and ethics documents.
Apex: Accountability. Institutions must ensure enforcement mechanisms exist: clear
sanctions for violations, defined appeals processes, and remediation (training, rewriting
assignments). Accountability also includes maintaining vendor oversight and auditing
high-risk applications.
By working up the pyramid, institutions create an environment where AI is used to support, not
undermine, scholarship.
Case Studies
Case Study A — Student Using AI for Assignments Without Attribution
Issue summary: A student submits an essay that contains fluent prose and advanced arguments
inconsistent with prior work. Several references are untraceable. The student used an LLM to
draft most of the essay and did not disclose this assistance.
Ethical analysis: The case constitutes misrepresentation of authorship and failure to meet
learning objectives. Even if the student made minor edits, the primary intellectual content was
produced by an AI, undermining assessment validity.
Legal implications: If the AI output includes text substantially derived from copyrighted
sources, this can implicate UU Hak Cipta. If fabricated claims defame individuals, UU ITE
may apply.
Recommended solution: Require resubmission with full disclosure, mandate an academic
integrity session, introduce process-based assessment components for the course, and
document the incident. If intentional deception is established, apply proportional sanctions per
university policy.
Case Study B — Lecturer Using AI to Evaluate Student Work Without Human
Oversight
Issue summary: A lecturer uses an AI tool to triage and assign preliminary grades for large
student cohorts, relying on automated feedback with minimal human review.
Ethical analysis: This workflow risks algorithmic bias, errors, and lack of pedagogical nuance.
Students are entitled to meaningful human feedback and recourse.
Legal implications: Uploading student data to third-party services may violate UU PDP if no
lawful basis or contract exists.
Recommended solution: Cease fully automated grading, ensure human-in-the-loop review for
final grades, obtain vendor contracts conforming to data-protection law, and provide
transparency to students about automation and appeal rights.
Case Study C — Misuse of Student Images to Make Deepfakes
Issue summary: A student creates deepfake videos of a fellow student as a prank and posts them
on social media.
Ethical analysis: The action violates personal dignity, consent, and privacy. It can cause
psychological harm and escalate into cyberbullying.
Legal implications: The deepfake may breach UU PDP for unauthorized processing of personal
data and UU ITE for dissemination of harmful electronic content.
Recommended solution: Immediate takedown, formal investigation, disciplinary sanctions,
offer support to the victim, require perpetrator to attend ethics training, and implement campus
wide policies preventing non-consensual media manipulation, including explicit consent forms
for public use of student images.
Practical Guidelines for Designing AI-Friendly Assignments
To preserve learning objectives while permitting ethical AI use, instructors should design tasks
that privilege process, reflection, and critical engagement. Examples include:
Scaffolded assignments: require initial outlines, annotated bibliographies, and draft
versions, which encourage iterative learning and allow instructors to monitor progress.
Prompt logs: students submit the prompts they used, the raw AI outputs, and a reflective
statement explaining how they transformed and verified the output.
Oral defenses or viva voce: require students to explain their reasoning and answer
probing questions, making it difficult to substitute AI for genuine understanding.
Peer review stages: students exchange drafts and perform structured peer assessment
focusing on argumentation quality and source use.
Assessment rubrics that reward process: allocate marks to evidence of critical
engagement and verification rather than solely polished final prose.
These strategies discourage misuse while harnessing AI as a scaffold that supports learning
rather than replacing it.
Verification & Attribution
Verification checklist:
Does every factual claim cite a primary or reputable secondary source?
Are all citations traceable via DOI, ISBN, or recognized databases?
Have you cross-checked disputed or surprising claims with two independent sources?
If code or data was generated, is there documentation of methodology and
reproducibility steps?
Attribution template (recommended):
This work incorporated assistance from [Tool Name, version] for [function(s) — e.g., idea
generation / language editing / summarization]. The author(s) revised and verified all outputs
and are solely responsible for the final content.
This template should be included in submissions, lab notebooks, and research methods sections
as appropriate.
Institutional Recommendations & Implementation - Campus-Wide AI Policy and Disclosure Standard. Draft a policy that defines
permissible AI uses, disclosure requirements, approved platforms, and the sanctions for
misuse. Create a standardized disclosure statement for submissions. - Mandatory AI & Data Protection Training. Implement compulsory short courses for
students and staff covering AI limitations, verification methods, prompt hygiene, data
protection, and legal obligations under UU PDP and UU ITE. - Approved Tools Registry & Vendor Contracts. Maintain a registry of vetted AI tools
with signed Data Processing Agreements that forbid model training on uploaded
academic data and specify retention and deletion policies. - Assessment Redesign & Academic Support. Provide templates and rubrics that value
process; fund writing centers and AI-literacy clinics to help students use tools
responsibly. - AI Ethics Oversight Committee. Form a multi-disciplinary committee to review high
risk AI applications, adjudicate incidents, and periodically revise policy in response to
technological change.
Implement these measures in phases: immediate policy adoption and training pilots (0–3
months), vendor contracting and assessment redesign (3–9 months), and full
institutionalization with monitoring and audits (9–18 months).
Assessment & Evaluation of the Module
To measure the module’s effectiveness, use a mixed-methods evaluation: pre/post surveys
measuring AI literacy and attitudes, audits of disclosure compliance in assignments, incident
tracking for policy violations, and qualitative feedback from focus groups of students and
faculty. Key performance indicators might include percentage of assignments with correct
disclosures, number of data incidents involving third-party tools, and improvements in
students’ ability to critique AI-generated content.
Module Summary
This mini-module provides a comprehensive and practical framework for the ethical, legal, and
responsible use of Artificial Intelligence within the academic environment of Universitas
Muhammadiyah Purwokerto. While AI offers substantial benefits—including enhanced
learning support, increased academic productivity, improved accessibility, and more efficient
research processes—its adoption must be balanced with a critical awareness of the risks
associated with misinformation, plagiarism, deepfakes, algorithmic bias, privacy violations,
and copyright infringement. By grounding the module in internationally recognized ethical
frameworks from UNESCO and the OECD, it emphasizes the importance of transparency,
fairness, accountability, safety, and respect for human dignity in all AI-related academic
activities. Furthermore, the module underscores that compliance with Indonesian legal
frameworks—namely the UU ITE, UU PDP, and Copyright Law—is essential to prevent
academic misconduct and potential legal consequences. Ultimately, it highlights that students,
lecturers, and researchers must practice responsible authorship, verify AI-generated
information, protect personal data, disclose AI assistance, and uphold academic integrity.
Supported by institutional policies, literacy programs, and monitoring mechanisms, UMP
can integrate AI in ways that strengthen its academic standards and promote ethical, human
centered use of emerging technologies.
References
Floridi , L., & Cowls, J. (2019). A Unified Framework of Five Principles for AI in Society.
Harvard Data Science Review.
Indonesia, P. P. (2008). Undang-undang (UU) Nomor 11 Tahun 2008 tentang Informasi dan
Transaksi Elektronik. Jakarta: Pemerintah Pusat.
Indonesia, P. P. (2014). Undang-undang (UU) Nomor 28 Tahun 2014 tentang Hak Cipta.
Jakarta: Pemerintah Pusat.
Indonesia, P. P. (2022). Undang-undang (UU) Nomor 27 Tahun 2022 tentang Perlindungan
Data Pribadi. Jakarta: Pemerintah Pusat.
Li, B., Qi, P., Liu, B., Di, S., Liu, J., Pei, J., . . . Zhou, B. (2023). Trustworthy AI: From
Principles to Practices. Association for Computing Machinery.
OECD. (2019). AI Principles. OECD.AI.
UNESCO. (2021). Draft text of the Recommendation on the Ethics of Artificial Intelligence.
France: UNESDOC.