AI features are now appearing in discovery systems, journal databases, citation indexes, legal platforms, e-book services, and reference managers. They do not all work in the same way. Some search a defined database, some answer from abstracts, and some can work with selected full-text documents.
This page is designed to answer four practical questions:
Can I use the feature through NCCU?
Do I need a personal account?
What content can the AI actually search?
What should I check before using the answer?
Primo Research Assistant turns a natural-language question into search terms, retrieves five records, and writes a short answer from their abstracts. You can ask in different languages, but the language of the question may change both the answer and the recommended records.
The five results are not necessarily the most relevant or most highly cited publications, and they are not a complete literature search.
If NCCU does not provide the full text of a recommended item, search the exact title in Google Scholar to look for an open-access version.
Access: NCCU Library login required
Source scope: selected electronic resources in Primo
Not included: physical books and audiovisual materials
During a search, ProQuest may suggest related terms.
Selected documents include a Research Assistant that can summarize the document, identify concepts and research topics, and suggest related resources.
EBSCOhost Academic Search Complete
EBSCO's natural-language search can help users enter a topic without first knowing the database syntax. Selected records may include AI-generated insights.
JSTOR's AI tools support semantic searching and questions about selected full-text items. The results-based chat works with the first 25 results and requires a personal account.
JSTOR is particularly valuable for humanities and social sciences, but it does not replace a comprehensive, up-to-date disciplinary database.
JSTOR's AI Research Tool, JSTOR Conversational Discovery GPT
The Oxford AI Discovery assistant accepts a question and recommends ten Oxford books or articles.
ACM provides AI summaries for many recent articles and longer summaries on selected article pages. Some conference pages may also include AI-generated podcast-style session summaries.
Citation databases normally provide titles, abstracts, references, and citation relationships. The article itself is usually hosted elsewhere.
Scopus AI is useful for entering an unfamiliar topic, identifying key papers and authors, visualizing concepts, and generating a Scopus query.
The AI Query Builder converts an English question into a database search. The standard AI response generally synthesizes a selection of Scopus abstracts and may show key publications, authors, a concept map, related questions, and possible emerging themes.
Deep Research creates a longer report and allows the user to specify conditions such as publication years, document type, or journal. A generated research gap should be treated as a question for further investigation, not as proof that no one has studied the topic.
A personal Scopus account can save conversations. Temporary chats and deletion options may also be available. If the system uses a source outside Scopus, check how that source is labeled.
Research AI answers questions using Statista reports and statistics, links back to the relevant material, and suggests related questions. A personal account is required.
The Ask AI feature answers questions about WRDS content.
AI-supported news and financial-statement summaries may be available on Bloomberg terminals.
WARC's question interface can respond by drawing on advertising and marketing content. A personal account is required.
Selected articles in the Law Journal Library display an AI-summary icon and a preview of roughly 300 words.
Register a personal account and use Beck Chat where it is available within NCCU's licensed modules. Verify the current law, jurisdiction, quotation, and procedural context.
ProQuest Ebook Central includes a Research Assistant in online reading mode for selected books.
iG Library includes AI search in its newer interface.
The Oxford English Dictionary AI Search Assistant supports question-based exploration of OED entries. Read the full entry when usage history or examples matter.
EndNote 2025 can extract information from PDFs attached to your EndNote library.
Writefull provides academic-English support in Word and Overleaf. Register with an NCCU email address. It can help with phrasing, abstracts, titles, and possible missing citations.
How should AI use be disclosed, and can software detect it?
For journal submission, follow the journal's AI policy and the appropriate guidance from APA, MLA, Chicago, or the relevant professional body.
AI-writing detectors can produce false positives and false negatives, and their mechanisms are often not transparent. Some institutions have discontinued the use of AI detectors because of the risk of unfair accusations.
Generated prose may show recurring vocabulary, punctuation, sentence patterns, or generic argument structure, and assignments that require sources may contain fabricated or misrepresented citations. These observations can prompt a closer review; they do not establish authorship.
Can Scopus AI conversations be saved?
A signed-in personal Scopus account can save conversations and show history. Users can also choose temporary chat or delete a thread.
Primo Research Assistant versus Scopus AI
Both accept questions in Chinese or English, but they draw from different collections.
Primo uses selected NCCU electronic resources and synthesizes the abstracts of its top five results.
Scopus AI uses Scopus-indexed records and offers stronger query building, concept maps, key authors and papers, themes, and Deep Research.
Primo may retrieve more Chinese material than Scopus, but neither is comprehensive. Both rely largely on abstracts and may expand or distort what a source says.
How can I avoid losing control of the search direction?
Write down the research question and scope before opening the AI tool. Ignore or correct suggestions that do not fit.
When the question is still developing, compare several tools or rephrase the same question to see which directions recur. Combine AI exploration with database filters, date and document-type limits, citation sorting, and manual browsing. These controls give the researcher more authority over what enters the evidence set.