What is Artificial Intelligence?
Artificial intelligence (AI) refers to the development of computer systems and machines that can perform tasks and exhibit behaviors typically associated with human intelligence. This includes the ability to think, learn, problem-solve, make judgments, and respond in ways that mimic or replicate human cognitive abilities (Frakenfield, 2022). AI technology is often used to describe any software or hardware that demonstrates human-like capacities for perception, reasoning, language understanding, and decision-making.
What is Generative Artificial Intelligence?
Generative artificial intelligence (AI) is an emerging and rapidly evolving technology. Similar to the internet as a whole, these AI tools, such as ChatGPT, Gemini or DALL-E, are not inherently good or bad when it comes to accessing and utilizing information. Rather, they represent a new paradigm in the way we engage with and interact with information.
Gratia Christian College’s AI policy
Gratia Christian College’s AI policy
General Guidelines
ChatGPT and Artificial Intelligence in Higher Education Quick Start Guide (2023, from UNESCO)
A Generative AI Primer by Michael Webb, 11 May 2021 (published by JISC)
AI Limitations
To learn more about why AI cannot always accurately mimic human behavior, this video below explores how AI does not usually have the nuances of human behavior and reasoning because they are focused on one thing: their purpose, which is why it often makes mistakes:
Check & Cross-check the Results
Many generative AI text tools have a common problem of producing nonsensical content, including fabricated citations or references.
When using text generated by ChatGPT or other AI assistants, it is important to carefully review and evaluate the information, just as you would with any content you plan to use for academic purposes.
It is crucial to use your own critical thinking and judgment to assess the quality and accuracy of the information, rather than blindly relying on AI-generated text.
Evaluating Information from Generative AI Tools:
1. Cross-check the information
- Look up the information on reliable sources like Wikipedia or Britannica
- Verify if other trustworthy sources corroborate the same information
- Compare the content against your own knowledge
- This practice of cross-checking, known as “lateral reading”, is a fundamental scholarly skill that predates the use of generative AI tools. To learn more about lateral reading, watch:
Lateral Reading (3 min 33 sec) by the University of Louisville Library
Check Yourself with Lateral Reading video (14 min) / by John Green (part of the Crash Course series)
2. Confirm the validity of any references or citations:
- Verify the provided references against databases like Library Catalogue or Google Scholar.
- Ensure that the references actually support the claims made in the AI-generated text.
3. Acknowledge the use of AI tools:
- If you incorporate content generated by ChatGPT or other AI assistants in your work, you should explicitly acknowledge this fact.
- Ensure that the references actually support the claims made in the AI-generated text.
The key takeaway is that when using text generated by AI tools, it is essential to apply the same rigorous evaluation and fact-checking process that is expected for any scholarly work, regardless of the source. This critical assessment is not a new requirement, but rather an established best practice in academic research and writing.
Research Help AI Tools
There are many new online tools being developed every day. Knowing which tool to use when preparing your assignment can definitely help make the process of putting together an assignment easier. The types of tool we are going to cover include:
Disclaimer: The Library do not subscribe to any of these tools. You are encouraged to evaluate the information on your own for validity and accuracy.
Semantic Scholar – A free, AI-powered research tool for scientific literature.
Elicit.org – Elicit uses language models to help you automate research workflows, like parts of literature review.
ChatPDF – Summarizes pdfs for you – how, who owns it? Not clear.
Consensus – Consensus is a search engine that uses AI to find insights in research papers.
AI Literacy – The Robot Test
Possessing artificial intelligence (AI) literacy does not mean you need to understand the advanced mechanisms of AI. It means you are actively learning about the technologies involved, and you will approach any text you read about AI, especially news articles, with a critical eye.
We have created a tool that you can use to help consider the legitimacy of the technology when reading about AI applications.
Reliability, Objective, Bias, Ownership, Type
Sources: Hervieux, S. & Wheatley, A. (2020). The ROBOT test [Evaluation tool]. The LibrAIry. https://thelibrairy.wordpress.com/2020/03/11/the-robot-test
Citing AI content in APA style
How to cite ChatGPT – APA Style Blog (updated on Feb 23, 2024)
In this post, APA Style Blog editors discuss situations where students and researchers might use ChatGPT or other AI to create text and to facilitate their research and how to cite it using APA Style.
Suggested practices:
1. Quoting or reproducing the text created by ChatGPT (or other AI models)
- Cite the content as software or algorithm’s output (rather than personal communication as there’s no actual person communicating)
- Describe how you used ChatGPT or other AI models in the Method or Introduction of your work
- Include the prompt used and any relevant generated text in your paper
- For long responses, consider including them in an appendix
In-text:
When prompted with “Is the left brain right brain divide real or a metaphor?” the ChatGPT-generated text indicated that although the two brain hemispheres are somewhat specialized, “the notation that people can be characterized as ‘left-brained’ or ‘right-brained’ is considered to be an oversimplification and a popular myth” (OpenAI, 2023).
2. Creating a reference to ChatGPT or other AI models and software
- Cite the model or tool as software
Anthropic. (2024). Claude-3.5-Sonnet [Large language model]. https://poe.com/Claude-3.5-Sonnet
Perplexity. (2024). Perplexity.ai [Large language model]. https://www.perplexity.ai/
OpenAI. (2023). ChatGPT (Mar 14 version) [Large language model]. https://chat.openai.com/chat
To read more about citing ChatGPT in APA Style, please see:
APA Citation Guide: AI – This guide from Seneca Libraries can help you cite AI in your assignments.
Comparison of GenAI Tools
(updated on Feb 19, 2025)
| Tool | Developed by | Type | Access | LLM model used | Advantages | Drawbacks | Privacy |
| ChatGPT | OpenAI | Chatbot | https://chat.openai.com/ (VPN required) | o3-mini,o1-mini,GPT-4o,GPT-4-turbo,GPT-4,GPT-3.5 | – (Jan 2025) o1-mini, o1-mini-high released – (Sep 2024) o1-mini, o1-preview, powerful models good at reasoning – (Jul 2024) GPT-4o-mini, a light weight model | – Advanced models have limited quota for free users – Not accessible in HK | User inputs and outputs may be used to train models unless you opt-out. https://openai.com/policies/privacy-policy “We may use Content you provide us to improve our Services, for example to train the models that power ChatGPT” |
| Poe | Quora | Chatbot | https://poe.com/ | DeepSeek-R1-Distill,DeepSeek-V3,o1-mini,o1-preview,GPT-4o-mini,GPT-4o,GPT-3.5,GPT-4,Claude 3.5,Claude 3,Claude 2,Claude 2.1,Claude 1,Genimi 1.5 Pro (Google),Gemini 1.0 Pro (Google),PaLM (Google),LLaMA (Meta),StableDiffusionXL,Mixtral,DeepSeek-R1 | – Accept PDF upload – Offers access to 100+ models from OpenAI, DeepSeek, Claude, Google, Meta, etc. – Share a chat in full or only a part – Customize your own bot | – Limited quota for o1 and other advanced models | https://poe.com/privacy “Keep in mind, any information and files you provide to the bots on Poe will be shared with third party AI model providers and developers powering the bots, and there is no need to share sensitive personal information with the bots (e.g., credit card information, social security information, etc.).” |
| Bing Chat (Copilot) | Microsoft | Chatbot,General search engine | https://copilot.microsoft.com/ | GPT-4,GPT-4-turbo,DALLE-3 | – Free GPT-4/GPT-4-turbo (at non-peak times) – Search Internet – Provide sources – Generate images (DALLE-3) | – Answers may include ads – Image generation feature (DALLE-3) requires VPN | User inputs and outputs may be used to train models. https://www.microsoft.com/en-us/bing/chat/enterprise/?form=MA13FV “User and business data in Copilot is protected and will not leak outside the organization.” https://learn.microsoft.com/en-gb/copilot/faq “Copilot with commercial data protection doesn’t retain your chat data and it isn’t used to train the underlying large language models.” |
| Gemini (Dec 2023) | Chatbot | https://gemini.google.com/ (VPN required); available through Poe.com | Gemini 2.0,Genimi 1.5 Pro (Google),Gemini 1.0 Pro (Google) | – Multimodal (trained on text, code, image, audio, video) – (Dec 2024) Gemini 2.0 Flash released – (Feb 2024) Gemini 1.5 up to 1 million token context window – Gemini Ultra model is reportedly more powerful than GPT-4 in tasks e.g. python coding, reading comprehension, math, etc. – (Feb 2024) Gemini Pro now powers Bard, and Gemini Nano will run on mobile devices. | – Gemini’s latest models are available on Gemini Advanced, which requires a subscription. | User inputs and outputs may be used to train models. Data where possible will be anonymised. Google have also reminded users not to include information that may identify the user. https://support.google.com/gemini/answer/13594961 “To help with quality and improve our products (such as generative machine-learning models that power Gemini Apps), human reviewers read, annotate, and process your Gemini Apps conversations.” “Please don’t enter confidential information in your conversations or any data you wouldn’t want a reviewer to see.” | |
| Claude | Anthropic | Chatbot | https://claude.ai/ (VPN required); available through Poe.com | Claude 3,Claude 3.5 | – (Jun 2024) Claude-3.5-Sonnet released – (Apr 2023) Claude-3 family released; Opus – most capable for reasoning, math and coding; Haiku – fast; Sonnet – a balance between skill and speed, still 2x faster than Claude-2 – (Claude-2 new in Nov 2023) Large context window (200k tokens) – Lower hallucination rate – A particular focus on safety (“Constitutional AI” – train AI with a set of principles) – Claude 2 is trained up to early 2023 | – Not accessible in HK – Currently 200k context window is reserved for Claude Pro users only | User inputs and outputs are not used to train the models unless you opt-in. However, prompts and conversations may be used to train where explicit permission has been given. https://www.anthropic.com/legal/privacy “We may process your personal data in an aggregated or de-identified form for the purposes of analysing the effectiveness of our Services, conducting research, analysing the behaviour of users of our Services and training our models and interfaces.” |
| Bard (rebranded to Gemini) | Chatbot,General search engine | https://gemini.google.com/ (VPN required) | Gemini 1.0 Pro (Google) | – Search Internet – Can further verify answer with Google – Offers extensions (Google Maps, Youtube, etc) | – Not accessible in HK | https://support.google.com/gemini/answer/13594961 “To help with quality and improve our products (such as generative machine-learning models that power Gemini Apps), human reviewers read, annotate, and process your Gemini Apps conversations.” “Please don’t enter confidential information in your conversations or any data you wouldn’t want a reviewer to see.” | |
| Perplexity | Perplexity | Chatbot,General search engine,Academic search tool | https://www.perplexity.ai/ | GPT-3.5,GPT-4 | – Search Internet – Provide sources – Accept PDF, CSV, image upload | – Limited quota for Pro Search (5 questions every 4 hours) – Limited quota for PDF, CSV, image upload | https://blog.perplexity.ai/legal/privacy-policy “You may opt out of data collection for AI data collection which would allows us to use your search data to improve our AI models in your settings page if you are logged into our platform.” |
| Google Search Generative Experience (SGE) (Nov 2023) | General search engine,Chatbot | https://www.google.com (VPN required) | PaLM 2 (Google),MUM | – Integrated with Google Search – Designed to be more like a search engine than a chatbot – More conservative in providing answers (than Bing Chat) | – Include academic sources | ||
| Grok (Aug 2024) | xAI | Chatbot | https://grok.x.ai/ (waitlist, currently limited to X verified users) | Grok-2 Beta | – Trained on X (twitter) data – Has real-time access to X (all tweets) – An “uncensored chatbot” – designed to answer “spicy questions” that are rejected by other AI systems | – Currently Limited to verified X users only | https://x.ai/legal/privacy-policy/ “To train our models, but only with the Usage Data you voluntarily choose to provide to us. For clarity, we do not train our models using your Account Information, Payment Information, Contact Details, or Technical Usage Data.” |
| Notion AI | Notion | Note-taking,Chatbot | https://www.notion.so/product/ai | “Models by Anthropic, OpenAI, Cohere” | – Generate/edit text directly in Notion (no need to copy-paste from a separate chat) – Customize column with tailored questions – (new in Nov 2023) “Q&A” allows users to ask questions about content in their own Notion | – Limit Free quota per month | https://www.notion.so/notion/Privacy-Policy-3468d120cf614d4c9014c09f6adc9091 |
| Elicit | Elicit | Academic search tool | https://elicit.com/ | GPT-3.5,GPT-4,Other open source models | – Accept PDF upload – Generate literature matrix – aim for high recall – Customize sources – Trained on Semantic Scholar and OA article full-text – Search articles by natural language query – Synthesize findings into a summary with real citations – (new in Jul 2024) Advanced Search, allows search by specific journal, author, DOI, OA status, citation count, etc. – (new in Jun 2024) Add tags to papers & filter papers by tags – (new in Mar 2024) Notebooks feature to run multiple queries, chat with papers, combine papers from search and uploads, and summarize papers – (new in Dec 2023) Import papers from Zotero – Export to CSV, RIS, BibTeX (Plus users only) | – One-time free quota: 5000 credits | https://elicit.com/operations/privacy |
| Consensus | Consensus | Academic search tool | https://consensus.app/search/ (Consensus GPT also available in GPT store) | GPT-4o,GPT-3.5,GPT-4 | – Customize sources – Trained on Semantic Scholar data – Search articles by natural language query – Synthesize findings into a summary with citations (Copilot feature) – Extract statements that support the claims – Label articles with study type, journal reputation, etc. – Export as CSV, RIS, BibTeX – Save search history | – Limited free quota (20 credits) for Synthesize and Copilot features | https://consensus.app/home/privacy-policy/ |
| Undermind | Undermind | Academic search tool | https://undermind.ai/ | GPT-4 | – Trained on Semantic Scholar data – Leverage LLM + citation mapping techniques – (Jun 2024 – limit to Pro users only) “Deep search” – retrieves around 100 papers and good at discovering papers by topics – Encourage detailed query (“explain your topic as you would to a colleague) – Synthesize findings into a summary with real citations – Group results by categories – Show level of matching for each result – (new in May 2024, limit to Pro users) View citation network with a timeline – (new in May 2024) Export as RIS, BibTeX | – Free tier: 10 searchers per month (show top 5 results only) | https://www.undermind.ai/privacy_policy/ |
| Petal | Petal | Academic search tool | https://www.petal.org/ | – Accept PDF upload – Generate literature matrix – Customize sources – Ask question based on selected sections in One PDF – Ask questions based on multiple PDFs | |||
| SciSpace | SciSpace | Academic search tool | https://typeset.io/ (SciSpace GPT also available in GPT store) | – Accept PDF upload – Generate literature matrix – Customize sources – Ask question based on selected sections in One PDF – Ask questions based on multiple PDFs | |||
| Scopus AI | Elsevier | Academic search tool | Through Scopus database | GPT-3.5,Other open source models | – Trained on Scopus data – Search articles by natural language query – Synthesize findings into a summary with citations from Scopus – Suggest topic experts (most productive authors in the field) – Show foundational papers (most cited relevant papers) – Generate a concept map to show connection between keywords – Export to CSV, RIS, BibTeX, Plain text | – Extra charge on top of Scopus subscription | https://www.elsevier.com/products/scopus/scopus-ai “Legal and technology protections to ensure zero data exchange or use of Elsevier data to train OpenAI’s public model” |
| Dimensions AI Assistant | Digital Science | Academic search tool | (Feb 2024) Dimensions Research GPT available in GPT store | – Trained on Dimensions data – (new in May 2024) GPTs are freely available for all ChatGPT users | https://www.dimensions.ai/privacy/ | ||
| GitMind | GitMind | Mindmapping | https://gitmind.com/ | – Text to mindmap – Expand ideas or questions based on One node | – Limited Free quota | ||
| Whimsical | Whimsical | Mindmapping | https://whimsical.com/ai/ai-text-to-flowchart | – Text to mindmap | – Limited Free quota | ||