Can You Use AI in HKDSE SBA and School Coursework?

If you are submitting a School-based Assessment (SBA) for the HKDSE or preparing an Internal Assessment (IA) and Extended Essay (EE) in Hong Kong, the short answer is yes, but with strict limitations and mandatory declarations. While generative AI tools like ChatGPT, Claude, and Copilot can serve as powerful brainstorming and coding assistants, presenting AI-generated text or data analysis as your own original work is classified as academic misconduct by both the Hong Kong Examinations and Assessment Authority (HKEAA) and international boards like the IBO.

To answer the primary question of how to cite AI in coursework: you must acknowledge the specific tool used, provide the exact prompt and date of generation, document the purpose of the output in an appendix or declaration form, and reference the model using standardized formatting (such as APA 7th, MLA 9th, or Harvard style). Submitting an assessment without declaring AI assistance risks marks deduction, disqualification of the entire SBA component, or overall grade downgrades.

Exam Board AI Compliance Matrix: What Is Allowed vs. Prohibited

Different examination bodies maintain distinct boundaries on permissible AI usage. Below is a breakdown of current expectations across secondary school assessment frameworks commonly undertaken by students in Hong Kong:

1. HKEAA (HKDSE School-Based Assessment)
• Permitted: Initial topic brainstorming, grammar check assistance, checking syntax for Information and Communication Technology (ICT) code, and gathering background literature for elective research projects.
• Prohibited: Generating actual essay body paragraphs for English Language Part B SBA, faking laboratory observation records in Biology/Chemistry/Physics SBA, or generating artwork rationales in Visual Arts without disclosure.
• Declaration Requirement: Students must sign standard HKEAA authenticity declaration forms confirming that all submitted materials represent their own independent synthesis.

2. International Baccalaureate (IBO - EE, TOK, and IAs)
• Permitted: Using AI as an interactive search engine, refining research questions, generating practice datasets, and formatting bibliography entries.
• Prohibited: Copy-pasting AI text into final essays without quotation marks and in-text citations.
• Declaration Requirement: Mandatory documentation in the Researcher's Reflection Space (RRS), Viva Voce discussions, and formal in-text referencing alongside a comprehensive prompt appendix.

3. UK Exam Boards (Pearson Edexcel & Cambridge International IAL)
• Permitted: Exploratory reading, scaffolding essay outlines, and vocabulary enrichment.
• Prohibited: Paraphrasing AI outputs to disguise machine-generated phrasing or submitting AI-produced solutions for Non-Exam Assessment (NEA) projects.
• Declaration Requirement: Compliance with the Joint Council for Qualifications (JCQ) AI guidance and signed candidate authentication forms.

Standard AI Citation Formats (APA, MLA, and Harvard)

When you reference content or ideas derived from an AI conversation, apply standard academic referencing conventions directly in your bibliography and footnote sections.

1. APA 7th Edition

APA treats generative AI as software creation and personal communication. When quoting or paraphrasing an AI conversation, structure your entry as follows:

In-text Citation: (OpenAI, 2024)
Reference List: OpenAI. (2024). ChatGPT (GPT-4o version) [Large language model]. https://chat.openai.com

2. MLA 9th Edition

MLA focuses on documenting the specific prompt entered by the researcher.

In-text Citation: ("Detailed prompt text" or shortened prompt)
Works Cited: "Explain the historical impact of the 1911 Revolution on Hong Kong commercial trade" prompt. ChatGPT, 24 Sep. version, OpenAI, 15 Oct. 2024, chat.openai.com.

3. Harvard System

In-text Citation: (OpenAI ChatGPT, 2024)
Bibliography: OpenAI (2024) ChatGPT (Version GPT-4o) [Large language model]. Available at: https://chatgpt.com (Accessed: 12 October 2024).

The Audit-Proof Prompt Log: Defending Against AI Detectors

School submission portals and teacher marking suites increasingly employ scanning tools such as Turnitin's AI Writing Indicator. However, AI detectors are known to produce false positives—flagging well-structured, formal academic English written by non-native speakers as machine-generated text.

To protect yourself against unjustified academic integrity investigations, maintain an Audit Trail while writing your SBA or coursework. Keep a dedicated digital folder containing the following five verification items:

1. Time-Stamped Version History: Always draft your essays in platforms that save incremental cloud revision histories (such as Google Docs or Word for Microsoft 365). A complete chronological edit history proves you typed and edited the essay over several days rather than pasting an entire essay at once.
2. The Raw Prompt Appendix: If you used AI for conceptual clarification, paste the complete conversation URL or export a PDF transcript of your chat log into an appendix folder.
3. Source Reading Notes: Keep handwritten notes, highlighted journal printouts, or digital bookmarks of your primary sources.
4. Outline Evolution: Save early sketches, mind maps, and initial drafts showing how your thesis evolved.
5. Revision Logs: If you utilize comprehensive study resources and notes to benchmark syllabus requirements, cross-reference your source points directly in your footnotes.

How to Handle a False AI Flag from Your School

If a teacher or school panel flags your HKDSE SBA or internal coursework draft for high AI probability, follow this four-step defense protocol:

Step 1: Request the Turnitin or Detector Report
Politely ask to inspect the exact diagnostic report to see which specific sentences and paragraphs were flagged.

Step 2: Present the Cloud Edit Timeline
Open the document's version history in front of your panel or teacher. Demonstrate the progression of keystrokes, revisions, and structural edits across multiple working sessions.

Step 3: Provide Primary Source Mapping
Show the physical textbooks, past papers, or scholarly articles from which your factual arguments were derived. Point out specific quotations and original data calculations that an AI model could not have fabricated without specific localized knowledge.

Step 4: Offer an Oral Defense (Viva Voce)
Offer to explain the line of reasoning, analytical frameworks, and terminology used in the flagged section verbally without referring to notes. Being able to explain your methodology clearly in your own words resolves the vast majority of false detection concerns.

Leveraging AI Ethically for HKDSE Preparation

Using AI tools responsibly is not about taking shortcuts—it is about accelerating your conceptual mastery. Rather than using automated models to write coursework, focus on deploying AI for targeted skills practice, such as debugging calculation steps, testing command word comprehension, and simulating mock exam questions.

Students looking for rigorous, syllabus-aligned test practice can leverage platforms like Thinka's AI-powered learning platform to identify revision blind spots, while educators can visit custom practice generation tools to build targeted question sets for their classes. You can also explore the interactive student practice platform to master exam techniques across core and elective subjects safely and transparently.