1. Primo Research Assistant (beta)

2. Scopus AI：In the Scopus database, click on the “Scopus AI” tab to ask questions using natural language. The AI will convert your question into keywords, select relevant high-quality literature (from 2003 onward) indexed in Scopus, and provide summarized answers in English. You can check the linked sources for full content and explore key articles, influential authors, concept maps, potential research topics, and suggested follow-up questions. If you have a personal Scopus account, you can save and review your chat history. For questions outside Scopus coverage, AI will still provide answers but will mark them as external results. The Copilot function also shows the keywords AI uses, which you can apply in other databases. For details, see Scopus's introduction site.

3. Statista：In the Statista database, click on the “Research AI” tab to ask questions using natural language. AI will provide answers ibased on Statista’s data and reports, with links to original sources and suggestions for related questions and resources. For more information, see Statista’s introduction page.

4. EndNote 2025：AI can automatically extract key points from full-text PDFs saved in your EndNote Library. This feature requires registration and login to your EndNote Web account. For details, see Endnote's instruction page.

5. Writefull：Register using your NCCU domain email (@nccu.edu.tw, @g.nccu.edu.tw, or @mail2.nccu.tw) and install the plugin. Writefull helps you edit English academic writing, check for missing citations, and generate titles, abstracts, and rewrites using Writefull X (AI features). See Writefull’s introduction.

6. ProQuest Central：AI suggests related keywords during searches. Some articles include the Research Assistant feature, which provides AI-generated summaries, recommended resources, related topics, and key concepts. Google’s AI summarizer can also help outline article structures. See ProQuest Research Assistant FAQs for more details.

7. ProQuest Ebook Central：When reading e-books online, you can use the Research Assistant feature in the left menu to view summaries of key points, important concepts, and recommended topics. See ProQuest Ebook Central Research Assistant FAQs for more information.

8. EBSCO (Academic Search Complete)：Supports natural language queries. Some documents also include the AI Insight feature, which provides quick summaries. See EBSCO AI function page for further details.

9. SciVal：A personal account is required to use SciVal. It analyzes researchers, institutions, and literature sets to identify popular research topics. AI generates topic summaries based on Scopus literature, helping you understand topic content and develop new research ideas.

10. Oxford Academic：In journal or book search pages, click the “AI Discovery Assistant” icon near the top right search bar to use the AI Q&A function. You can ask questions in Chinese, and AI will reply in English with answers sourced from Oxford books and journals. It suggests up to 10 articles but does not provide summarized answers. 

11. Oxford English Dictionary：Click the “AI Search Assistant” at the top right corner to use the AI Q&A function. 

12. iG Library：The new interface now includes AI search functions.

AI Functions Requiring Self-Application for Access

• JSTOR：See the page for instructions on how to apply for AI query access (untested; please try it if interested). JSTOR is part of ITHAKA, a non-profit organization that also compiles and compares various emerging AI tools.

Databases with AI Functions Requiring Additional Payment – Please Utilize Available Functions

• ScienceDirect: See Generative AI on ScienceDirect. You can use “Ask ScienceDirect AI” for natural language queries. For empirical studies, AI can suggest research designs. The Reading Assistant summarizes articles.

• FT.com: See the AI function introduction.

• Westlaw: See the AI function introduction.

• Web of Science: See the AI function introduction. NCCU subscribes to the Social Science Citation Index only. 

• WARC: See the AI function introduction and watch workshop videos for other features.

• Turnitin: See the AI function introduction. Note: Current AI detectors have error rates, and some universities choose not to subscribe to avoid misjudgments.

To write effective prompts, you can refer to the CLEAR framework, OpenAI Prompt Engineering resources, the Prompt Engineering Guide , the University of Sydney's AI in Education (which provides prompt examples across disciplines).

🔸 Provide sufficient information and specific requirements: The clearer you are about your goals and the more background information you provide, the better your results will be. Include any known facts, ideas, and specify requirements such as word count, writing style, or table formats.

🔸 Follow up: Break down complex questions into smaller ones, rephrase or ask from different perspectives, request AI to ignore previous instructions and answer afresh, ask for the reasoning behind its answers, challenge its responses, request further elaboration, ask for pros and cons, or encourage it to think outside the box.

🔸 Use examples: Explain the kind of content you want by giving examples, or ask AI to provide examples to illustrate its explanations.

🔸 Set roles: Define the role you or the AI is playing, the context, and the task to be completed. Providing clear context helps AI generate more relevant responses.

🔸 Avoid bias: Do not include obvious biases or preferences in your prompt, as this affects the output. Ask AI to provide sources to support its responses when needed.

🔸 Confirm understanding: Ensure AI correctly understands the information you provide before asking it to proceed. Initially, assign tasks you are familiar with so you can better judge its accuracy.

🔸 Use dedicated chat threads for specific tasks: Create a separate conversation thread for each task, such as translation or summarization, and continue asking related prompts within that thread to build a more effective workflow.

🔸 Verify information across multiple sources: Always cross-check information obtained from AI with other sources, then return to ChatGPT for further clarification or exploration as needed.

Learning and Planning

🔺 You can ask AI for advice on time management, group task distribution, progress tracking, academic writing styles, and formatting.

🔺 Try asking AI to act as your teacher to review and revise your report or thesis. To do this effectively, first gather your professor’s grading rubric or examples of excellent reports/papers, and ask AI to edit your work based on these standards so you can learn areas for improvement.

Brainstorming, Exploring Topics, and Focusing Your Thesis

🔺 AI can provide basic structures, suggestions, and help clarify your ideas by identifying gaps in your thinking. The more knowledge and background you provide, the better your prompts and results will be. Challenge AI’s responses to co-create better outcomes.

🔺 Research topics often come from unique life experiences or social observations and combining knowledge from different fields, including both popular and niche areas. A good topic should stand out, be feasible, and address a gap in the literature while aligning with your interests. Your initial observations from daily life or reading journals can be supplemented by AI’s large training dataset, but AI cannot replace your own reading and critical thinking.

🔺 When you doubt your topic, AI can provide encouragement and affirmation, helping you take the first step in writing.

Literature Searching

🔺 When using AI tools to search for literature, check each tool’s coverage. For example, Scopus AI searches English literature from 2003 onward indexed in Scopus. Many articles require subscriptions, so AI search results do not represent all available literature. Always use multiple search tools to obtain comprehensive results.

🔺 If ChatGPT or other large language models suggest literature, always verify each article title using Google Scholar or library databases to confirm its existence.

🔺 Even when AI provides real articles, they may not be the most relevant or high-quality sources, and AI search results are usually limited. Use academic AI tools with filtering and sorting features, treat AI results as starting points, and extract keywords from AI-recommended articles for further searches. You can also use citation snowballing to find related articles.

🔺 Be cautious if the AI tool lacks training data in specific languages, cultures, or historical periods; its answers in these areas may be unreliable. Always fact-check AI outputs, especially for high-stakes decisions in fields like medicine, law, or finance.

🔺 You can also ask AI to recommend databases and search keywords.

Reading, Summarizing, and Reference Management

🔺 Do not upload subscription-based articles downloaded from library databases to external AI platforms (unless labeled open access). Before using any AI tool, check its privacy and copyright policies to avoid unintentional privacy leaks or copyright violations.

🔺 AI can explain unfamiliar terms, but for in-depth understanding, consult additional academic sources.

🔺 AI is useful for quickly identifying whether an article contains perspectives relevant to your report or thesis, especially when you already have an outline and know what information is needed for each section.

🔺 Be aware of AI summarization limitations. Many AI tools have input size limits and perform better on shorter segments. Avoid directly copying AI summaries or translations into your assignments or thesis.

🔺 You can ask AI for tutorials on proper summarizing, paraphrasing, and citing. While many academic AI tools generate basic citation formats, you should always double-check them for accuracy.

🔺 For proper AI citation formats, refer to Wayne State University Libguide and citation styles like APA, MLA, or Chicago. Most academic publishers do not recognize AI as an author since it cannot take responsibility for its outputs. Check sources yourself rather than citing AI-generated statements directly. If submitting to journals, mention AI use in your methods or acknowledgements, ensure human review and editing, and note that reproducibility may be an issue since AI outputs can vary with the same prompts.

Research Methods, Data Collection, and Analysis

🔺 For detailed research design, it is still better to consult theses and journal articles. However, AI can provide suggestions when you are stuck in designing your methods or remind you of key considerations mentioned in methodology textbooks. You can ask AI to check for potential improvements in your research design.

🔺 Some AI tools can transcribe interview recordings and analyze text, transcripts, surveys, or other research data. Search online for user experiences and be mindful of data privacy risks when using such tools.

🔺 During data analysis, always relate AI outputs to your research questions and literature review to ensure consistency. Avoid relying solely on AI analyses without integrating them into your thesis narrative.

Writing, Editing, and Translation

🔺 Provide AI with high-quality evaluation criteria for writing or translation, and ask it to revise your text while explaining its changes.

🔺 Avoid directly using AI-generated text. Current AI-written outputs may include repetitive structures or expressions influenced by training data.

🔺 Regardless of AI assistance, have experts proofread your writing.

🔺 Ultimately, you as the author are responsible for your work and cannot blame AI for any errors.

1. Search Tools

• Perplexity: Allows you to customize search scopes for each question, such as limiting to academic sources or including social media and video content for tool reviews or user experiences. It quickly displays source links for fact-checking and offers a deep research function.

• Felo

• Deep research functions across different models: This feature allows you to explore a topic in greater depth by providing comprehensive summaries, related keywords, and connections to other relevant articles or concepts, helping you gain a deeper understanding beyond surface-level search results.

🔹 Although these AI tools can access real-time online information and provide sources, they may misinterpret or emphasize/deemphasize content incorrectly. Always check the original source to confirm its meaning and context, as AI outputs can carry biases.

2. Academic Literature Exploration

🔹 Useful for citation chaining, finding related articles beyond initial keywords, and visualizing articles by publication year and citation count.

• Litmap

• Connected Papers

• ResearchRabbit

🔹 Most AI literature search tools source articles from open-access databases like Semantic Scholar or OpenAlex, which do not cover all disciplines, books, statistics, or grey literature. Many resources remain accessible only via subscription databases or physical collections. For government statistics or institutional reports, checking official websites is faster and more accurate than asking AI.

3. Comprehensive Academic Platforms

🔹Integrating literature search, reading assistance, writing support, and citation generation

• SciSpace: Supports natural language literature queries with sortable results and partial summaries. Allows file uploads for AI-assisted reading and interpretation. Includes writing, rewriting, AI content detection, and citation formatting features. Recent updates allow research topic exploration and PDF-to-video conversion.

• Elicit: Supports natural language queries with sortable results and partial summaries. Allows file uploads for AI-assisted reading and interpretation. Offers strong research topic exploration and deep research features.

• Paperguide: Similar to SciSpace and Elicit, with natural language querying, sortable results, partial summaries, file upload for reading assistance, writing and rewriting support, AI content detection, and citation formatting.

• Paper Digest: Focused on computer science. Offers natural language literature queries with sortable results, partial summaries, and deep research features, as well as writing and rewriting support.

4. Literature Review

• Undermind: Unlike other search tools, Undermind first asks clarifying questions to narrow your research focus before presenting literature results organized by topic concept and chronological order. It takes longer to generate reports and has usage limits in the free version.

5. Literature Organization and Visualization

• Notebook LM: Supports organizing various data types with literature conversation functions.

• Napkin AI: Recommends suitable charts based on text input.

• Gamma AI: Generates presentation slides from text or documents and creates accompanying images based on slide content.

🔹Consider exploring productivity, writing, and research GPTs in the GPT Store. Before using, you can always ask each GPT for instructions on optimal usage. The GPT canvas function is also worth exploring.

🔹Aside from the tools listed above, many AI tools exist for writing, visualization, qualitative and quantitative data analysis, image generation, presentation creation, and transcription. It is recommended to research user reviews before registering and to test free versions to evaluate performance before considering paid plans. 

🔹Choosing AI Tools: The ROBOT Test

• Reliability: Is the developer trustworthy? How transparent are they? Could their tools carry bias?

• Objective: What are the developer’s motivations and goals? Might these affect how you use the tool?

• Bias: What potential biases or ethical issues might arise from this tool?

• Owner: Who owns the technology—company, government, academic institution, or individual? Does ownership affect accessibility?

• Type: What type of tool is it, and in what contexts is it appropriate to use?

🔹Reflective Questions Before Using AI Tools

• Why do I want to use AI in this context? Will it reduce my learning opportunities? How can I use it to achieve good outcomes while enhancing my learning?

• What data was this AI tool trained on? What functions does it have, and why is it suitable for my current task?

• Could this AI tool introduce bias? Does providing it with my data risk violating my or others’ privacy or copyright? How can I avoid these issues?

• What prior knowledge do I have about this topic? Is it enough to evaluate AI outputs, or should I use other methods or tools for cross-checking?

• Embrace an experimental mindset: Test AI outputs using topics you already know well to understand its strengths and limitations. Could using other information sources or different AI tools yield better or faster results?

💠 Users cannot fully prevent AI hallucinations; AI generates results based on probabilities and is unaware of its own errors.

💠 You can reduce inaccuracies by requesting sources, asking AI to self-check, challenging its answers, or providing correct information for it to base further outputs on.

💠 Always remain critical and verify AI-generated information, just as you would double-check a speaker’s accidental mistakes in a lecture. The more familiar you are with a topic, the easier it is to notice errors. Continue building reliable knowledge from diverse sources rather than over-relying on AI.

💠 When you already know where to find accurate data (e.g., official government statistics sites), use those instead of AI, which may require extra time to verify. For tasks with clear correct answers, such as citation formatting, AI can assist but should not replace manual checking.

💠 Remember that AI output quality depends on its training data. Consider whether the AI tool has access to the data you need, especially if it is subscription-based or restricted content.

💠 Even when using AI to generate tables or translations based on correct input data, always review the output for potential errors.

💠 Please refer to the SIFT method and the CRAAP test for information verification. These approaches are applicable to all types of information evaluation.

💠 Be sure to consult with your course instructor or thesis advisor regarding how you use AI, and always keep a record of your AI usage to document which ideas in your report or thesis came from AI and which are your own. You can refer to University of Cambridge's template to indicate how you used AI in your work.

💠 For a comprehensive overview of AI-assisted writing, please refer to Monash University’s guidelines.

It is only available in Chinese, but you may use AI to translate it. 

The most important part states:

• Under the regulations set by your course instructors and in compliance with academic ethics, students may appropriately use generative AI tools to enhance their learning efficiency and outcomes, and to complete course assignments and projects.

• Students should have a clear understanding of both the advantages and disadvantages of generative AI tools, as well as the potential risks and intellectual property issues involved in their use. For example, AI-generated content may contain errors or significant biases, or it may include works protected by others’ intellectual property rights. Users must critically assess, thoroughly review, verify, and revise AI outputs, including their wording, and ensure that intellectual property rights are properly cited.

• When using generative tools to produce content for reports or assignments, students must rigorously cite sources to meet academic integrity requirements.