Skip to Main Content

Generative AI

5 ways to use ChatGPT for Research

  1. Brainstorm & define your research  - a means of dialogue to identify correlations & gaps
  2. Summarizing & translating academic articles
  3. Review research before submitting for peer review - good for proofreading
  4. Polish academic writing - grammar & style 
  5. Interrogate data with some generated code

Resources

 
Consensus
 https://consensus.app/search/
Uses AI to find answers in research papers. Can ask questions about relations between concepts, a yes/no question or the effects of a concept GPT 4  • Familiar search interface
• Can perform searches without signing up for account
• Easily share results with others
• Provides brief summary of paper, journal details where it was found and link to full text
• Can filter results by publication date and study type 
• Works best if phrase is asked in question form
• Limited scope - covers economics, health supplements, medicine, mental health, sleep and social policy
• Unable to locate any information about what their dataset is (where they get their academic article from)
• "This journal is in the top 5% of our database" - however we don't know how large database is 
• Free option: Unlimited search and research quality indicators. Three GPT-4 summaries (600 word input) and consensus meters a month
•Premium [USD 9.99 p/m]: Unlimited search, research quality indicators, GPT-4 summaries (1,200 word input) and consensus. Bookmarks coming soon
•Enterprise [Custom price on application]: Premium + manage organisational accounts, share bookmarks and lists with your team and integrate with your research library
• 40% student discount available (unsure if Aus supported) 
https://consensus.app/results/?q=Is%20agile%20an%20effective%20project%20management%20style%3F&synthesize=on
Elict

https://elicit.org/
Uses language models to help you automate research workflows, like parts of a literature review. Can find relevant papers without perfect keyword match, summarize takeaways from the paper specific to your question, and extract key information from the papers. "language models like GPT-3"

castorini/monot5-base-msmarco-10k
• Provides details about dataset used (115M papers from the Semantic Scholar Academic Graph dataset)
•  Uses semantic similarity, which finds papers related to your question even if they don't use the same keywords
• For every search result, Elicit reads the abstract and generates a custom summary that is relevant to your question
• "80- 90% accurate, definitely not 100% accurate."
• Unable to see past FAQ'S without signing up 
• Unable find any information without going through signup process 
scite.ai

https://scite.ai/
scite uses access to full-text articles and its deep learning model to tell you, for a given publication:
• how many times it was cited by others
• how it was cited by others by displaying the text where the citation happened from each citing paper
• whether each citation offers supporting or contrasting evidence of the cited claims in the publication of interest, or simply mention it
Not advised  • Zotero and Chrome plugins
• Detailed help guides
• Large dataset with some information about data included
• Can flag citations for review
Ability to make custom dashboards using Zotero & Mendeley libraries, manual comma-separated listed of DOIs or CSVfiles. 
• No summary generated, but creates a smart citation
• Creating dashboards can be time consuming
• 7 day day trial
• Monthly AUD $22.21
•Yearly AUD 159.94
Very popular for evaluting sources 
Research Rabbit

https://www.researchrabbit.ai/ 
ResearchRabbit is a “citation-based literature mapping tool”. The
scope of such tool is to optimise your time searching for references as you start planning your
essay, minor project, or literature review.
Not advised  • Users can add papers to 'collections', the AI then learns what you like and provides you with tailored recommendations.
• Can represent collections and connections between them visually
• Allows searching within the interface.
• Can import and export from citation managers
• dataset not disclosed
• has a higher learning curve due to the variety of options available
Free  Singapore Management University has a very good guide: https://library.smu.edu.sg/topics-insights/new-literature-mapping-tool-researchrabbit
Chat PDF

https://www.chatpdf.com/
Your PDF AI - like ChatGPT but for PDFs. Summarize and answer questions for free. Chat GPT-3.5 • Users upload PDF and can then ask questions, create summaries and tests.  • Can only look at a few paragraphs from the PDF at once
• doesn't understand images and has issues with tables
• Can only 'chat' with one pdf at a time
• Free: 3 PDFs every day, each up to 120 pages, 50 questions per day.
• USD 5 per month : 50 PDFs every day, each up to 2000 pages, 100 questions per day.
Perplexity

https://www.perplexity.ai
AI search engine with powerful academic research abilities. Chat GPT-3.5, GPT-4 if paid • Connected to the internet
• iPhone app
• Focus allows you to fine tune your search by narrowing down the sources, for more targeted and relevant results.
• Copilot allows you to ask clarifying questions to guide the AI search process and find the best answer 
• Struggles with PDF's  • Free option: limited access to GPT-4
•Premium: USD 20 per month. Access to GPT-4, dedicated support, 
Semantic Scholar

https://www.semanticscholar.org/
An academic search engine driven by artificial intelligence that enables users to filter through millions of scholarly pieces.   •  Covers disciplines including biology, medicine, computer science, geography, business, history, and economics
• No login required
• Can export library 
• Limited research scope
• evaluation metric are narrow 
Free 
Subjects: Business / Advertising, marketing and public relations, Business / General business, Business / Management and human resources
Tags: bibliographic management, bsn503, business journals, carma, citation tracking, data analysis, literature review, research, research effectiveness, researcher impact, scholarly publishing, stata