GenAI Research Assistants

Depending on the service and feature, a generative AI research assistant may:

• search for papers using a question rather than keywords alone
• generate an answer based on papers it has retrieved
• summarise or compare several papers
• extract information such as methods, participants, findings or limitations
• organise extracted information into a table
• answer questions about a selected paper or collection
• support parts of screening or data extraction for a literature review

Some tools search their own collection of academic records. Others may also use available full text, search external sources or analyse documents supplied by the user.

A generated answer or table is an interpretation of the source material. It is not a replacement for reading and evaluating the original papers.

Always check your assessment brief. It will explain whether generative AI can be used for the task and whether your use must be acknowledged.

Abertay’s supported general-purpose AI assistant for students is Microsoft Copilot Chat. The specialist research assistants listed below are external services. Abertay does not require you to use them, and you do not need to pay for one to complete your studies.

Features, account requirements, usage limits and terms can change. Check the provider’s current information before registering or entering any material.

For guidance on deciding whether and how to use AI in research, see AI for researching a topic.

These tools increasingly provide overlapping features. The examples below illustrate different ways in which AI can be combined with academic search and analysis. Inclusion on this page does not mean that Abertay endorses or provides the service.

Elicit

Elicit searches scientific literature using natural-language research questions. It can display papers in structured tables, extract information from them and generate research reports.

It also provides features intended to support parts of literature-review work, including screening and data extraction. These may be useful for investigating a focused research question or comparing studies, but extracted information must be checked against the original papers.

Consensus

Consensus searches academic literature before generating an answer or synthesis. It can summarise and compare studies, extract details such as methods and sample sizes, and show patterns of agreement or disagreement for some questions.

It may be useful for gaining an initial view of research relating to a clearly framed question. The resulting synthesis depends on the papers retrieved and how the tool interprets them, so it should not be treated as a settled account of the evidence.

scite

scite combines academic searching and an AI assistant with information about how publications have been cited by later research.

Its Smart Citations feature classifies citation statements as supporting, contrasting or mentioning the cited work. This may help you investigate how a publication has been discussed, but an automated classification does not determine the overall quality of a paper or establish whether its conclusions are correct.

Scholar Labs 

Scholar Labs is an experimental AI-powered feature within Google Scholar. It analyses a detailed research question, searches Google Scholar and generates brief explanations of how selected papers relate to the question.

Its main purpose is finding and evaluating the relevance of individual papers rather than producing a complete synthesis across the literature. Availability may vary and a Google account is required.

Learn more in the Library’s Google Scholar guide.

Different tools and features may work from:

• publication titles and bibliographic information
• abstracts
• extracts from papers
• full text available to the provider
• documents uploaded or selected by the user
• citation statements taken from later publications

Do not assume that a tool has read the full text of every paper it cites. Its search may cover only part of the relevant literature, and its answer may be based on a smaller selection of papers than the complete results suggest.

Coverage can vary by subject, language, publication date and type of material. Books, chapters, professional sources, creative works and other material that matters in some disciplines may not be represented adequately.

A paper appearing in the results does not mean that its full text is available through the tool. Use LibrarySearch to look for access through Abertay or for an openly available copy.

Citations and links help you investigate an answer, but they do not prove it is accurate.

Before relying on an answer or table:

1. Confirm the papers exist and read the originals. Check that they are suitable for your task and support the claims attached to them.
2. Compare important extracted details, such as quotations, methods, sample sizes and findings, with the originals.
3. Consider whether the answer misses relevant or conflicting research, makes unsuitable comparisons or draws conclusions more certain than the evidence allows.

Use your subject knowledge to judge the papers and the tool’s interpretation. For the full method of checking sources, claims and synthesis, see AI for researching a topic.

Use a research assistant alongside your reading list, LibrarySearch and relevant subject databases. 

Check what sources or collections it searched and which searching or selection decisions you have delegated to it. Review any criteria it uses to include or exclude papers. Review the process as well as the final answer. If it does not provide the search details or selection criteria your project requires, do not assume that its process meets those requirements.

 Record the tool, question, date and relevant settings if your assessment or research method requires this. Acknowledge your use of AI as your assessment brief requires.

Some providers offer features described as supporting systematic reviews. Using one of these features does not by itself make a search comprehensive, reproducible or suitable for a systematic review. Follow the methodology required by your discipline, supervisor and review protocol.

Generative AI research assistants are usually question-led. They retrieve or analyse papers and generate an answer, summary, comparison or table.

Literature mapping tools are mainly relationship-led. You usually begin with one or more papers and explore citation, author or similarity connections around them.

Some services now include both approaches, but the distinction can still help you choose a starting point:

• use a research assistant to explore a question or compare information from papers
• use a literature mapping tool to explore connections around papers you have already identified

Learn more about literature mapping tools.

Do not enter personal, confidential or sensitive information into an external AI service. Do not upload:

• participant or research data
• unpublished research, findings or grant material
• substantial parts of an assessment, dissertation or thesis
• confidential professional or placement information
• journal articles, book chapters or other licensed material unless you have permission

These restrictions still matter if a tool invites you to upload complete documents.

See AI and data privacy for full guidance.