ACRL Artificial Intelligence (AI) Interest Group

Hi all,

What AI tools do you like for searching or exploring Scholarly Literature? 

Nicole's response Nov 8 on in the Discussion thread about things people don't know or are curious to learn more about Generative AI or AI in general seemed promising as a new discussion thread, so starting that here. :-) There are many other topics in the earlier thread that would be great to delve into more as well. Nicole also linked to a blog post on some of these tools, "There is a good article here about some of them: AI Tools for Research: a Sandbox Session Recap by Brian Mathews."

For one additional tool that I'm aware of:

Consensus is a scholarly literature searching / exploration AI tool that a Library colleague at Virginia Tech has pointed out to me as one that provides some amount of transparency right away as you use it. (Note, I can't really compare it with other tools yet as I haven't really used others for scholarly lit searching yet. As with all AI tools, I am still in the exploration phase for this one.) As a first pass, some things that seem helpful about it (the free version) are: it provides a brief summary at the top in response to a query, then provides several key points pulled from several (~10 usually it seems) articles, and each point has numeric citations to the references these more focused highlights are pulled from. Below the highlights is the reference list for the articles the highlights were pulled from and at a glance you can see the Title, one or more generated / summarized key Findings from the article, the Year, # of Citations, the First Author, the abbreviated Journal / Source title, at times an article type - at least if it's a Literature Review or Systematic Review, etc.  You can continue to click on 'More' at the end of each 10 article result set for additional results, though no total number of results is displayed at any time (that I can see). More info on How Consensus Works is on their site.  Per this page, for 'What does Consensus search,' it says, "The current source material used in Consensus comes from the Semantic Scholar database (About Semantic Scholar), which includes over 200 million papers across all domains of science. We continue to add more data over time and our dataset is updated monthly." It seems to allow saving lists of searches and papers you want to remember, and has info on how to integrate it with citation managers (Zotero, etc.), and more info on how the 'Consensus Meter' and other features work in their help documentation.

What are your thoughts on AI tools focused on searching or exploring scholarly literature? 

~Ginny

------------------------------
Virginia Pannabecker
Assistant Dean and Director, Research Collaboration and Engagement
University Libraries, Virginia Tech
She/Her/Hers,They/Them/Theirs
vpannabe@vt.edu
------------------------------

Hi Ginny and All,

I really like Keenious for research exploration.  It does not summarize things for you or do any writing for you.  What it does do is take phrases, paragraphs, or entire papers that you've written or a PDF of an article that someone has written and use AI to analyze the text and the context around the text to find articles to support those concepts.  Essentially, it is an alternate way to find articles when you've exhausted your ideas for keywords.  It gets its articles from OpenAlex.  You can try Keenious out for free but it is really geared for use at colleges and universities so they offer a subscription and will link the articles found to your library catalog so it also keeps students and faculty within your library system.  The other one that I like is Elicit.  It is similar to Consensus but early on when I did some testing of accuracy between the two, Elicit was more accurate in the sources that it cited in the summaries that it created.  Consensus may have improved since then.  Lastly, another fun one is Open Knowledge Maps. It is completely free.  No need to create an account.  It is a visual way of looking at articles.  It creates a map of a research topic to help find documents and identify relevant concepts from the keywords that you give it.  It obtains its papers from PubMed and BASE.

Thanks for starting this thread!  I'm interested to see what else is out there that you all like.

Kathleen

Kathleen Makarewicz, MD, MLIS

Science Librarian

Occidental College Library

Los Angeles, CA

------------------------------
Kathleen Makarewicz
Science Librarian
------------------------------

Hi everyone,

New to the group! I've come up with a process I've found useful for discovering new literature.

I started by exporting my Zotero library to PDF, including the abstracts and my notes with each citation. I then take that document and put it into NotebookLM, which is a Google AI product. NotebookLM allows you to load up to 50 documents at 500,000 words for each document into a particular environment, and then you can chat or create outlines, etc, even have it produce a podcast about the document(s) you upload. 

I then ask questions looking for certain documents (eg, I'm looking for articles in this document that use natural language processing methods on non-fiction texts like news articles). It will answer those prompts and provide links back to the Zotero output so you can review the citation closer and see if it is correct.

Then, I take some of those relevant articles and drop the DOI into something like LitMaps or Research Rabbit to find other similar articles. Then I can add those articles into my Zotero and build it up. 

Thanks for starting the thread! Interested to hear about the tools other people are using. 

David Arredondo, MLIS

Collections Librarian

Calvin T Ryan Library

University of Nebraska at Kearney

------------------------------
David Arredondo
Collections Librarian
University of Nebraska at Kearney

arredondodr@unk.edu
He/Him/His
------------------------------

Hello David, 

Thank you for your post. It was helpful and fun!

------------------------------
Eddie Lagos
Student
San Jose State University School of Library & Information Science
------------------------------

Original Message:
Sent: Nov 20, 2024 10:27 AM
From: David Arredondo
Subject: What AI tools do you like for Scholarly Literature Searching or Exploration?

Hi everyone,

New to the group! I've come up with a process I've found useful for discovering new literature.

I started by exporting my Zotero library to PDF, including the abstracts and my notes with each citation. I then take that document and put it into NotebookLM, which is a Google AI product. NotebookLM allows you to load up to 50 documents at 500,000 words for each document into a particular environment, and then you can chat or create outlines, etc, even have it produce a podcast about the document(s) you upload. 

I then ask questions looking for certain documents (eg, I'm looking for articles in this document that use natural language processing methods on non-fiction texts like news articles). It will answer those prompts and provide links back to the Zotero output so you can review the citation closer and see if it is correct.

Then, I take some of those relevant articles and drop the DOI into something like LitMaps or Research Rabbit to find other similar articles. Then I can add those articles into my Zotero and build it up. 

Thanks for starting the thread! Interested to hear about the tools other people are using. 

David Arredondo, MLIS

Collections Librarian

Calvin T Ryan Library

University of Nebraska at Kearney

------------------------------
David Arredondo
Collections Librarian
University of Nebraska at Kearney

arredondodr@unk.edu
He/Him/His
------------------------------

Here's an interesting article I came across recently about these types of tools (by Aaron Tay). http://musingsaboutlibrarianship.blogspot.com/2024/11/why-use-of-new-ai-enhanced-tools-that.html 

One things I liked is how he categorizes these tools as 1- answer engines, 2 - writing tools, and 3 - citation-based mapping of papers.

So I made this list (I hope it's correct, since I haven't tried all of these). It just helps me to have some kind of grouping when thinking about them.

1. Answer engines for academic search
* Elicit
* Consensus
* Scite AI
* Scopus AI
* Primo Research Assistant
* Web of Science Research Assistant

Answer engines for web search:
* Perplexity
* You.com
* Bing Chat
* ChatGPT Plus with search

2. AI writing tools

* Keenious
* SciSpace AI Writer
* Jenna.ai
* GPT 4o with Canvas

3. Citation-based mapping of papers
(Recommend papers based on your seed paper)

* Connected Papers
* Research Rabbit
* Litmaps

Hope this is helpful!

Nicole

------------------------------
Nicole Hennig
ELearning Developer
University of Arizona Libraries
nhennig@arizona.edu
------------------------------

That is a great article by Aaron Tay.  Thanks for sharing Nicole!  I do disagree with one of the categorizations though.  Keenious does no writing or suggesting at all.  All it does is a semantic search with text that you give it, either that you've written yourself or from an article.  The output is a list of relevant scholarly articles.  It functions as an alternative search method to keywords. 

Kathleen

------------------------------
Kathleen Makarewicz
Science Librarian
------------------------------

Original Message:
Sent: Nov 21, 2024 04:11 PM
From: Nicole Hennig
Subject: What AI tools do you like for Scholarly Literature Searching or Exploration?

Here's an interesting article I came across recently about these types of tools (by Aaron Tay). http://musingsaboutlibrarianship.blogspot.com/2024/11/why-use-of-new-ai-enhanced-tools-that.html 

One things I liked is how he categorizes these tools as 1- answer engines, 2 - writing tools, and 3 - citation-based mapping of papers.

So I made this list (I hope it's correct, since I haven't tried all of these). It just helps me to have some kind of grouping when thinking about them.

1. Answer engines for academic search
* Elicit
* Consensus
* Scite AI
* Scopus AI
* Primo Research Assistant
* Web of Science Research Assistant

Answer engines for web search:
* Perplexity
* You.com
* Bing Chat
* ChatGPT Plus with search

2. AI writing tools

* Keenious
* SciSpace AI Writer
* Jenna.ai
* GPT 4o with Canvas

3. Citation-based mapping of papers
(Recommend papers based on your seed paper)

* Connected Papers
* Research Rabbit
* Litmaps

Hope this is helpful!

Nicole

------------------------------
Nicole Hennig
ELearning Developer
University of Arizona Libraries
nhennig@arizona.edu
------------------------------

Original Message:
Sent: 11/22/2024 10:57:00 AM
From: Kathleen Makarewicz
Subject: RE: What AI tools do you like for Scholarly Literature Searching or Exploration?

That is a great article by Aaron Tay.  Thanks for sharing Nicole!  I do disagree with one of the categorizations though.  Keenious does no writing or suggesting at all.  All it does is a semantic search with text that you give it, either that you've written yourself or from an article.  The output is a list of relevant scholarly articles.  It functions as an alternative search method to keywords. 

Kathleen

------------------------------
Kathleen Makarewicz
Science Librarian
------------------------------

Thanks Kathleen for the clarification on Keenius and that it doesn't automatically write or add citations but merely suggestg based on the text you enter/highlight.

That said the name "AI writing tool" is a bit misleading on my part but what I was getting at is a class of tools that suggest citations as you write (See context of Use with caution! How automated citation recommendation tools may distort science  ), which is a big use case for Keenius given the affordances of the UI, 

 I understand of course, Keenius is just feeding the text into a semantic type search and giving a list of results, but the outcome is it may encourages users to find confirming evidence on the fly. 

You could of course try to do it with a "answer engine" but the link is less direct.

------------------------------
Aaron Tay
Librarian
National University of Singapore
------------------------------

Original Message:
Sent: Nov 22, 2024 09:57 AM
From: Kathleen Makarewicz
Subject: What AI tools do you like for Scholarly Literature Searching or Exploration?

That is a great article by Aaron Tay.  Thanks for sharing Nicole!  I do disagree with one of the categorizations though.  Keenious does no writing or suggesting at all.  All it does is a semantic search with text that you give it, either that you've written yourself or from an article.  The output is a list of relevant scholarly articles.  It functions as an alternative search method to keywords. 

Kathleen

------------------------------
Kathleen Makarewicz
Science Librarian
------------------------------


Original Message:
Sent: Nov 21, 2024 04:11 PM
From: Nicole Hennig
Subject: What AI tools do you like for Scholarly Literature Searching or Exploration?

Here's an interesting article I came across recently about these types of tools (by Aaron Tay). http://musingsaboutlibrarianship.blogspot.com/2024/11/why-use-of-new-ai-enhanced-tools-that.html 

One things I liked is how he categorizes these tools as 1- answer engines, 2 - writing tools, and 3 - citation-based mapping of papers.

So I made this list (I hope it's correct, since I haven't tried all of these). It just helps me to have some kind of grouping when thinking about them.

1. Answer engines for academic search
* Elicit
* Consensus
* Scite AI
* Scopus AI
* Primo Research Assistant
* Web of Science Research Assistant

Answer engines for web search:
* Perplexity
* You.com
* Bing Chat
* ChatGPT Plus with search

2. AI writing tools

* Keenious
* SciSpace AI Writer
* Jenna.ai
* GPT 4o with Canvas

3. Citation-based mapping of papers
(Recommend papers based on your seed paper)

* Connected Papers
* Research Rabbit
* Litmaps

Hope this is helpful!

Nicole

------------------------------
Nicole Hennig
ELearning Developer
University of Arizona Libraries
nhennig@arizona.edu

This is a really helpful breakdown, Nicole! I love how you've organized the tools into clear categories, which makes it easier to navigate and understand their different uses. I agree that having a structure like this is great for thinking about how each tool can serve a specific purpose whether it's for academic search, writing, or citation mapping. Thanks for sharing this list, it's definitely a useful reference!

------------------------------
Abigail Calderon
------------------------------

Original Message:
Sent: Nov 21, 2024 04:11 PM
From: Nicole Hennig
Subject: What AI tools do you like for Scholarly Literature Searching or Exploration?

Here's an interesting article I came across recently about these types of tools (by Aaron Tay). http://musingsaboutlibrarianship.blogspot.com/2024/11/why-use-of-new-ai-enhanced-tools-that.html 

One things I liked is how he categorizes these tools as 1- answer engines, 2 - writing tools, and 3 - citation-based mapping of papers.

So I made this list (I hope it's correct, since I haven't tried all of these). It just helps me to have some kind of grouping when thinking about them.

1. Answer engines for academic search
* Elicit
* Consensus
* Scite AI
* Scopus AI
* Primo Research Assistant
* Web of Science Research Assistant

Answer engines for web search:
* Perplexity
* You.com
* Bing Chat
* ChatGPT Plus with search

2. AI writing tools

* Keenious
* SciSpace AI Writer
* Jenna.ai
* GPT 4o with Canvas

3. Citation-based mapping of papers
(Recommend papers based on your seed paper)

* Connected Papers
* Research Rabbit
* Litmaps

Hope this is helpful!

Nicole

------------------------------
Nicole Hennig
ELearning Developer
University of Arizona Libraries
nhennig@arizona.edu
------------------------------

Hi all, 

This is Aster from Hong Kong University of Science and Technology library. My first post here. :) 

At my institution, I teach a workshop "Emerging AI tools for literature review", where I gave a broad overview of how GenAI tools can support different stages of a literature review and their limitations. Here are the tools I cover: https://libguides.hkust.edu.hk/AI-tools-literature-review/ (overview on homepage, a comparison of tools under 2nd tab, and the workshop slides under last tab). 

For literature discovery in particular, my personal take would be Scite, Elicit, and more recently Undermind.ai. All three are trained on Semantic Scholar data and provide real sources. The key features I like are: 

• Scite extracts citation statement which makes easier for users to verify. 
• Elicit generates a giant literature matrix which is good for extensive review (and in my experience, the articles it retrieved are more relevant than other similar tools).
• Undermind is designed for extensive review too, suggests 100 articles in the first run (though some articles are particially hidden for free users). I like their way of guiding users to refine their question, make it more explicit, before proceed with searching. I learned this tool from Aaron Tay's post, and as I understand that their library has subscribed to this tool. 

Hope these info are helpful. Happy to discuss more. 

Best,
Aster

------------------------------
Aster Zhao
Research Support Librarian
Hong Kong University of Science and Technology
------------------------------

Thanks, that is useful! I will read your article later today.

------------------------------
Nicole Hennig
ELearning Developer
University of Arizona Libraries
nhennig@arizona.edu
------------------------------

Original Message:
Sent: Dec 09, 2024 12:09 AM
From: Aster Zhao
Subject: What AI tools do you like for Scholarly Literature Searching or Exploration?

Hi all, 

This is Aster from Hong Kong University of Science and Technology library. My first post here. :) 

At my institution, I teach a workshop "Emerging AI tools for literature review", where I gave a broad overview of how GenAI tools can support different stages of a literature review and their limitations. Here are the tools I cover: https://libguides.hkust.edu.hk/AI-tools-literature-review/ (overview on homepage, a comparison of tools under 2nd tab, and the workshop slides under last tab). 

For literature discovery in particular, my personal take would be Scite, Elicit, and more recently Undermind.ai. All three are trained on Semantic Scholar data and provide real sources. The key features I like are: 

• Scite extracts citation statement which makes easier for users to verify. 
• Elicit generates a giant literature matrix which is good for extensive review (and in my experience, the articles it retrieved are more relevant than other similar tools).
• Undermind is designed for extensive review too, suggests 100 articles in the first run (though some articles are particially hidden for free users). I like their way of guiding users to refine their question, make it more explicit, before proceed with searching. I learned this tool from Aaron Tay's post, and as I understand that their library has subscribed to this tool. 

Hope these info are helpful. Happy to discuss more. 

Best,
Aster

------------------------------
Aster Zhao
Research Support Librarian
Hong Kong University of Science and Technology
------------------------------

Oh, it's a Libguide, great! I love the infographic on the first page.

------------------------------
Nicole Hennig
ELearning Developer
University of Arizona Libraries
nhennig@arizona.edu
------------------------------

Original Message:
Sent: Dec 09, 2024 10:58 AM
From: Nicole Hennig
Subject: What AI tools do you like for Scholarly Literature Searching or Exploration?

Thanks, that is useful! I will read your article later today.

------------------------------
Nicole Hennig
ELearning Developer
University of Arizona Libraries
nhennig@arizona.edu
------------------------------


Original Message:
Sent: Dec 09, 2024 12:09 AM
From: Aster Zhao
Subject: What AI tools do you like for Scholarly Literature Searching or Exploration?

Hi all, 

This is Aster from Hong Kong University of Science and Technology library. My first post here. :) 

At my institution, I teach a workshop "Emerging AI tools for literature review", where I gave a broad overview of how GenAI tools can support different stages of a literature review and their limitations. Here are the tools I cover: https://libguides.hkust.edu.hk/AI-tools-literature-review/ (overview on homepage, a comparison of tools under 2nd tab, and the workshop slides under last tab). 

For literature discovery in particular, my personal take would be Scite, Elicit, and more recently Undermind.ai. All three are trained on Semantic Scholar data and provide real sources. The key features I like are: 

• Scite extracts citation statement which makes easier for users to verify. 
• Elicit generates a giant literature matrix which is good for extensive review (and in my experience, the articles it retrieved are more relevant than other similar tools).
• Undermind is designed for extensive review too, suggests 100 articles in the first run (though some articles are particially hidden for free users). I like their way of guiding users to refine their question, make it more explicit, before proceed with searching. I learned this tool from Aaron Tay's post, and as I understand that their library has subscribed to this tool. 

Hope these info are helpful. Happy to discuss more. 

Best,
Aster

------------------------------
Aster Zhao
Research Support Librarian
Hong Kong University of Science and Technology