Monday, 1 December 2025

A note on Harvard referencing practice on AI-generated contents: for MBA and Housing Studies students

A note on Harvard referencing practice on AI-generated contents: for MBA and Housing Studies students (information from perplexity.ai)


My question: "With my MBA and Housing Studies students increasingly using AI tools (notably perplexity.ai) to inform them to do assignments and dissertation projects, please provide 5 illustrations of how to make citations on AI-generated contents as well as how to provide references on them in Harvard referencing format."

The reply: "Students can cite AI tools in Harvard style by treating them either as personal communications, software, or as a specific “response to a prompt”, depending on local university guidance. Below are 5 practical illustrations you can adapt for MBA and Housing Studies work, including Perplexity.ai.​

General points

  • Always follow your own university’s official Harvard guide first, as many now publish specific AI examples.​

  • Good practice is to: briefly describe the prompt in the reference, include the AI tool name, developer, date of interaction, and (optionally) the URL or model/version.​

Illustration 1: AI used for background explanation (ChatGPT)

Use case: Student used ChatGPT to explain “capitalisation rates in real estate investment”.

In‑text citation

  • Paraphrase:
    Capitalisation rates can be interpreted as the expected annual return on a property’s current market value, before financing and tax (OpenAI ChatGPT, 2025).

Reference list entry (Harvard)

  • OpenAI ChatGPT (2025) ChatGPT response to prompt “Explain capitalisation rates in real estate investment for MBA students”, 1 March.

This format (Name of AI, year, description of response, date) is consistent with several university Harvard examples that treat AI as a distinct communication type.​

Illustration 2: AI used for literature scoping (Perplexity.ai)

Use case: Student used Perplexity.ai to identify key themes on “Hong Kong Gen Z housing affordability” before conducting a formal database search.

In‑text citation

  • Paraphrase:
    Preliminary themes included delayed household formation, a preference for renting, and cross‑border mobility (Perplexity.ai, 2025).

Reference list entry (Harvard, software/output style)

  • Perplexity.ai (2025) Perplexity.ai response to prompt “Summarise main themes in research on Hong Kong Generation Z housing affordability”, 10 March.

Some Harvard variants allow or encourage treating generative AI outputs similarly to software or algorithmic outputs, naming the tool as author.​

Illustration 3: AI used for methodological advice

Use case: Student asked an AI tool for advice on sampling strategies for semi‑structured interviews.

In‑text citation

  • Narrative:
    Advice on theoretical and purposive sampling was initially obtained from an AI assistant, which highlighted the need for diversity across income and tenure status (OpenAI ChatGPT, 2025).

Reference list entry (Harvard, with developer and URL)

  • OpenAI ChatGPT (2025) ChatGPT response to prompt “Recommend sampling strategies for semi‑structured interviews on Hong Kong Gen Z housing affordability”. OpenAI. Available at: https://chat.openai.com/ (Accessed: 5 March 2025).

Several institutional Harvard guides suggest including developer and access details when treating AI as software or an online tool.​

Illustration 4: AI used for draft text checking (Perplexity.ai)

Use case: Student pasted a draft paragraph and asked Perplexity.ai to suggest clearer phrasing (not to generate content from scratch).

In‑text citation (methods section)

  • The draft literature review was grammar‑checked and clarified with the assistance of an AI tool, without adding new substantive content (Perplexity.ai, 2025).

Reference list entry (Harvard, tool‑level reference)

  • Perplexity.ai (2023) Perplexity.ai (AI chatbot) [Large language model]. Available at: https://www.perplexity.ai/ (Accessed: 5 March 2025).

This mirrors examples where the AI platform itself is referenced as software rather than a specific conversation.​

Illustration 5: AI used to generate a table or list of factors

Use case: Student asked ChatGPT to list “key drivers of housing affordability in Hong Kong” and then critically evaluated each factor using peer‑reviewed literature.

In‑text citation

  • Paraphrase:
    An initial AI‑generated list of affordability drivers included income levels, housing supply constraints, mortgage policy, and intergenerational transfers (OpenAI ChatGPT, 2025).

Reference list entry (Harvard, “response to prompt” + institution style)

  • OpenAI ChatGPT (2025) Response to prompt “List key drivers of housing affordability in Hong Kong”, 15 March.

Several university Harvard exemplars recommend enclosing the prompt in quotation marks and using a descriptive phrase such as “response to prompt” for AI‑generated text.​

Comparison of 3 common Harvard-style approaches

Approach focus

Typical in-text form

Typical reference list pattern

Treat AI as specific response

(OpenAI ChatGPT, 2025)

OpenAI ChatGPT (2025)ChatGPT response to prompt "…", Day Month.

Treat AI as software/tool

(Perplexity.ai, 2025)

Perplexity.ai (2023) Perplexity.ai (AI chatbot)[Large language model]. URL (Accessed: date).

Hybrid (response + provider)

(ChatGPT, 2025) or (OpenAI ChatGPT, 2025)

ChatGPT, response to "Prompt used", OpenAI, Date. URL (optional).



These patterns reflect current Harvard‑style institutional guidance, which remains provisional and may differ by university, so students should always align with their programme handbook or library guidance.



Posted by at No comments:
Subscribe to: Comments (Atom)