10 Chapter 7: Reading the Literature

“All research begins from reading and understanding the ‘authorities on the subject’.” (Sivakumar & Lukose, 2017, p. 123)

Once you’ve completed your literature search and compiled a solid set of sources, the next crucial step is to read. And by read, we mean truly engage with the texts. We always recommend doing at least one full reading of each paper without AI assistance. This initial engagement helps develop an understanding of the structure, main arguments, methodology, and overall contribution of the work. Once you’ve done that, you can turn to AI tools to assist with deepening your comprehension such as seeking further clarifications, summarizing complex sections, or drawing connections across texts.

As we mentioned in the previous chapter, a literature review isn’t just a summary, it’s a critical evaluation of the work you’ve read. And critique sits high on the cognitive ladder. You can’t critically evaluate something you don’t understand. That’s why your own engagement with the material is non-negotiable.

So, what might an AI-enhanced reading process actually look like in practice? Here’s the strategy we suggest. Once you’ve gathered your materials, begin by reading each one yourself. As you go, use your preferred note-taking method to capture key ideas, quotes, arguments, counterarguments, whatever stands out.

After you’ve completed your first reading, or even a second or third, you might still have questions. Maybe a term wasn’t clearly defined, a method seemed overly complex, or a certain result left you puzzled. This is where AI tools can help. Platforms like NotebookLM, ChatGPT, Claude, and others mentioned later in this chapter can help you interact with the text more dynamically.

For example, let’s say you’re reading a study on teacher feedback in online learning, and you hit a wall when you get to the statistical analysis. You could paste that section into ChatGPT and ask, “Can you explain what a mixed-effects model is in plain language?” Or if you’re struggling to connect a paragraph to the paper’s overall thesis, you might ask, “How does this section support the author’s main argument?” These kinds of interactions won’t replace your own thinking but they will support it. AI can clarify, simplify, rephrase, or suggest related research, all of which help you build a more grounded and nuanced understanding of the work you’re reading.

Because you’ve already done that initial, unaided reading, you’re now much more familiar with the content of the paper. That puts you in a stronger position to evaluate AI-generated responses and spot any inaccuracies or hallucinations. This is exactly why the first reading matters, it grounds you in the material and gives you the context needed to judge the reliability of any AI support that follows.

To take full advantage of what AI tools can offer during the reading process, you’ll often need to upload the materials you’re working with. Doing so allows the AI to respond directly to your questions based on the actual content of the paper. However, this brings us to a critical point: the ethical handling of copyrighted material. Some AI platforms may use uploaded documents to further train their models, raising concerns about privacy and intellectual property rights.

That’s why it’s essential to check how each platform handles user data before uploading any documents. Look into their privacy policies, data retention practices, and user agreements. Most of the tools we recommend in this chapter are designed with privacy in mind, but it’s still important that you make informed decisions based on your comfort level and institutional policies.

With that in mind, let’s now take a closer look at the AI tools that can support your reading and analysis of literature. We begin with NotebookLM, a free and powerful platform that we highly recommend. In the next section, we walk you through how it works and how you can use it to build deeper, more interactive engagement with the texts you’re reading.

NotebookLM

NotebookLM is one of our favorite AI-powered tools for reading academic literature. Built by Google, NotebookLM is specifically designed to help you engage with the materials you upload. What makes it so valuable, especially for research, is that it doesn’t pull information from across the internet. Instead, it only uses the sources you upload to answer your questions. This focused, source-specific interaction is exactly what we want when doing close reading.

You can upload a variety of content to NotebookLM: PDFs, Google Docs, Google Slides, even YouTube URLs. And since its initial launch, the platform has evolved significantly. As of now, while we are writing this chapter, NotebookLM added new cool features like mind map generation visual graphs, and video overviews.

To get started using NotebookLM, log into your NotebookLM account and create a new notebook, think of it as a folder where you store all the reading materials for a specific project. Each notebook can hold up to 50 sources and you can create up to 100 notebooks. But here’s a tip based on our experience: don’t overload a single notebook. The chatbot can get confused if too many papers are uploaded at once. We usually upload about 10 to 15 papers per notebook for better performance and clearer responses.

Let’s say you’re working on a literature review. You could name your first notebook ‘Literature Review Part 1’, upload your first batch of 10–15 papers, and when you’re ready, create Part 2, Part 3, and so on. This way, you maintain clarity and avoid overwhelming the tool with too much information at once.

Once your documents are uploaded, NotebookLM takes you to a three-column interface (subject to change, of course, as features evolve). Here’s how it’s structured:

• The left column shows the list of all the documents you’ve uploaded.
• The middle column displays an automatically generated summary of the selected sources.
• The right column is for the Audio Overview, a relatively new and exciting feature we’ll talk about in a moment.

By default, all papers are selected after upload, meaning any question you ask in the chat will draw from all selected sources. But you can easily toggle this if you want to ask about a specific paper, just select it and click. This is great when you want to drill into particular sections like the methodology, findings, or theoretical framework. You can ask something like, “What research design did the authors use in this study?” or “Summarize the discussion section for me.”

NotebookLM also offers extra features to further enhance your reading. These include a Study Guide feature, which automatically generates structured summaries and highlights key ideas. Then there’s the FAQ feature, which creates a list of questions and answers based on the uploaded documents, another fantastic way to process dense material quickly.

When we interact with a paper in NotebookLM, we often save useful AI responses by clicking the ‘Save to Note’ button located above the chat box. This allows us to revisit insights without regenerating them. Later, we usually export these notes into our own note-taking system (e.g., Obsidian or Notion) where we can tag them, organize them, and connect them with other ideas in our research workflow.

NotebookLM also offers an Audio Overview which simulates a conversation between two AI hosts summarizing key topics from your uploaded sources. What’s important to note is that these discussions aren’t general commentaries, they’re grounded in the actual content you’ve uploaded. Even better, you can customize the conversation to focus on a particular topic, author, or intended audience.

It takes a few minutes for the overview to generate, and while we wouldn’t rely on this feature for complex academic tasks like literature reviews, it’s a fantastic option for lighter content. For example, we’ve used it to generate summaries of online articles, which we then listen to while driving or working out. That said, for critical reading and writing (like crafting a literature review) we prefer to stay focused and work directly with the papers on oure laptop. Still, it’s worth exploring, especially for revisiting material or reviewing more general sources.

Overall, NotebookLM will help you with your deep reading and with managing your research materials. Suppose you’re writing a draft and can’t recall which author discussed a particular theory. Just open your relevant notebook, select all papers, and ask, “Which author talks about X?” The AI will pull a response directly from your sources and best of all, it provides references so you can verify the information.

In short, NotebookLM allows you to centralize your academic materials, interact with them intelligently, and get help extracting meaning from dense texts all without losing track of where the information came from. It’s a powerful co-pilot for the reading phase of your research, and definitely a tool worth incorporating into your workflow.

AI Chatbots: Your Swiss Army Knife for Reading Academic Literature

We like to think of AI chatbots as the Swiss Army knife of AI. They’re versatile, adaptable, and surprisingly powerful when used well especially in the context of reading academic literature. These tools can help you streamline the entire process: you can upload your documents, ask questions, clarify complex concepts, generate summaries, identify research gaps, create self-assessment questions, and much more.

When we talk about AI chatbots here, we are mainly referring to the popular ones like ChatGPT, Claude, Perplexity, Gemini, and Microsoft Copilot. For the purposes of reading and engaging with academic material, we’ll focus on ChatGPT and Claude, as they both offer a particularly useful feature: Projects.

What Are Projects and Why Do They Matter?

Projects are essentially dedicated, self-contained workspaces. Each project has its own chat history and a personal knowledge base made up of the files you upload. It’s a bit like the notebook system in NotebookLM. When you use Projects in ChatGPT or Claude, you’re creating a custom environment where the chatbot draws only from the documents you’ve provided. It won’t pull from the web or its general training data, which drastically reduces the risk of hallucinated or irrelevant responses. This setup is ideal for deep academic reading. You’re not just chatting with a general-purpose AI, you’re working with a focused assistant trained only on your chosen materials.

Projects also give you consistency and continuity. Everything you discuss is saved within that workspace, and when you return later, the chatbot remembers what you’ve been working on and continues within that context. That’s a big plus when you’re juggling multiple documents or building a long-term research project. One thing to note: Projects are currently only available to paid users on both platforms.

Currently, both ChatGPT and Claude allow you to upload up to 20 files per project, which, if you recall from the NotebookLM section, is a good upper limit. We’ve found that uploading 10–15 papers per project strikes a nice balance between depth and performance. Too many documents can overwhelm the model and muddy the responses.

If you’re working with more than 20 papers, you might need to create multiple projects. For example, you could name your first one ‘Literature Review Part 1’, upload the first batch of papers, then create Part 2, and so on.

Another really helpful feature is the ability to add custom instructions to your project. You can guide the chatbot’s behavior by telling it how to respond, what tone to use, what role to take (e.g., act as an academic writing coach), or even what perspective to adopt (e.g., focus on educational theory or qualitative methods). This level of customization can make your conversations much more aligned with your goals.

Once your documents are uploaded and your custom instructions are in place, you’re ready to dive in. Just like with NotebookLM, you can ask your chatbot to explain complicated terms or concepts, summarize key findings, identify overarching themes, compare studies, or even generate questions to test your understanding.

However, there’s one limitation to keep in mind: unlike NotebookLM, ChatGPT and Claude don’t let you select individual papers to interact with inside a project. All uploaded documents are considered together when the chatbot generates responses. That means if you ask a question, it pulls from the whole pool of files.

We’ve tried workarounds like including the title and author in the prompt to focus the chatbot on a specific paper but the results were hit or miss. If we want to engage deeply with a single paper, we usually skip the Projects feature altogether. Instead, we upload one paper in a fresh chat and work with it there. That way, all interactions are specific to that document, and we can easily save the chat history or copy the content into my note-taking system like Obsidian or Notion. So, we typically use Projects for grouped readings and switch to single uploads when we want to do a close reading of one paper at a time.

Think of Projects as your space for taking a bird’s-eye view. Once you’ve done the foundational work (i.e., reading, annotating, and digesting each paper) you can turn to Projects to ask bigger, more integrative questions. For example, you can ask the chatbot to compare how different authors discuss a particular theme, identify patterns of argument or recurring theoretical frameworks, or even trace the evolution of an idea across several studies. This kind of cross-referencing is incredibly valuable when you’re trying to weave a coherent narrative out of diverse sources.

Another great way to use Projects is for self-assessment. You can prompt the chatbot to quiz you based on the content of your uploaded papers. For instance, you might ask: “Can you generate five questions that test my understanding of the methodology sections in these papers?” or “What are some key differences in how these studies define formative assessment?” This interactive Q&A approach helps reinforce your learning and uncover gaps in your understanding.

Projects is also ideal when you’re exploring research gaps or limitations in your field. Since the chatbot has access to all the documents you’ve uploaded, you can ask it to highlight areas that remain underexplored or to summarize how various authors frame the limitations of their studies. This can provide a valuable springboard for your own research focus, helping you identify promising avenues to investigate further.

And as we said earlier, none of this works well without a solid note-taking system running alongside your reading. To turn passive reading into active, engaged learning, you need to externalize your thinking. Personally, I (first author) always keep our Obsidian workspace open while I interact with papers in Projects. As I work, I copy and paste useful quotes, paraphrase key insights, record themes, note contradictions, and draft mini-reflections. Then I go back to the chatbot, ask follow-up questions, clarify doubts, and refine my understanding.

It’s an iterative process: reading sparks questions, questions lead to new insights, and those insights deepen your grasp of the literature. Over time, your notes become a rich, personalized map of the knowledge you’re building. And that’s where the magic really happens: when AI supports your critical thinking without replacing it.

Perplexity AI Spaces

As we are writing this book, we come back to this chapter to add this new feature called Spaces that has been recently released by Perplexity AI. It is much like Projects in ChatGPT and Claude: a dedicated, persistent workspace where you can upload files, organize your threads, and interact with your sources in a focused, ongoing way.

Spaces allow you to create a knowledge hub for a specific research topic. You can name the Space, set custom instructions for how the assistant should respond, and upload up to 50 documents if you’re on the Pro plan.

Spaces offers much flexibility in how you query your content. You can choose to search only your uploaded files, any added links, or combine those with live web results for a broader inquiry. You can also attach files temporarily to a specific thread without saving them to the Space, useful for one-off explorations.

Another feature worth noting is collaboration. If you’re on the Pro plan, you can invite up to five collaborators to work within the same Space. Everyone in the group can view and interact with the uploaded materials, making it a useful option for joint research efforts or advising scenarios.

Now, aside from NotebookLM, ChatGPT, Claude, and Perplexity AI Spaces, there are several other AI-powered tools that can support you in reading and engaging with academic literature. Some of these tools were already introduced in the previous chapter when we looked at AI for literature search, while others appear here for the first time. In what follows, we give you a quick overview of each one so you can explore and experiment with the ones that best suit your workflow and preferences.

Consensus

We’ve already discussed Consensus in the previous chapter as a great tool for literature search, but it’s also quite helpful when it comes to reading the literature itself. One of its most useful features is the ability to upload individual papers and interact with them directly. All your uploaded documents are stored in what’s called the ResearchHub in your Consensus account.

Once a paper is uploaded, you can use the “Ask this paper” feature to query it. You might ask for a summary, clarification of a complex concept, or an explanation of a figure or chart. Consensus will not only provide an answer but also highlight the exact section of the paper where that information is found. You can dig into the methodology, data analysis, or even ask for key findings which makes it a handy tool for close reading. That said, the platform currently allows you to chat with only one paper at a time, so it’s more useful for focused reading than for synthesis across multiple sources.

Elicit

Elicit takes a different approach and is especially helpful for those looking to engage with multiple papers at once. For premium users, Elicit allows you to chat with up to eight full-text papers simultaneously. You can search for papers directly in the Elicit database or upload your own PDFs including imports from platforms like Zotero. Once you’ve selected the papers you want to explore, you can enter a chat mode where you can ask broader questions like “Compare and contrast the findings across these studies” or more targeted queries depending on your focus.

The interface shows whichh papers are being referenced in responses, and you can choose to use either the full text (if available) or just the abstracts, which speeds things up and reduces processing costs. This makes Elicit especially useful when you’re trying to identify patterns, align findings, or build a comparative understanding of a specific topic. It’s a solid option for synthesizing material and gaining perspective across several studies at once.

R Discovery

R Discovery is another valuable AI tool that can support your literature reading process in multiple ways. Like other tools we’ve discussed, R Discovery allows you to upload full-text research papers and ask questions about their content. The chatbot can help clarify concepts, summarize key points, or direct you to specific sections relevant to your query.

One particularly helpful feature is its ability to generate instant audio summaries or even read the entire paper aloud. This can be incredibly useful if you want to reinforce your first or second reading by listening to the content while commuting, walking, or doing chores. Turning a dense academic article into an audio format offers a convenient way to revisit material and deepen comprehension on the go. While it may not replace focused, note-taking sessions, it’s a great supplemental tool for busy researchers who want to stay immersed in their material.

SciSummary

SciSummary offers a rich set of features designed to enhance your engagement with academic papers. It uses AI to generate summaries for the documents you upload, but it goes beyond that. The platform includes an interactive chat function that allows you to engage with papers on a deeper level. You can ask follow-up questions, request clarifications, and even explore visual elements like figures.

For example, when a paper includes charts or images, SciSummary gives you access to a list of all figures through a dedicated “Figures” tab. You can click on any title to get a breakdown of what the figure means, or submit your own question using the built-in prompt box. This makes it especially helpful for interpreting complex data visuals that might otherwise slow down your reading process.

What’s more, SciSummary allows you to generate summaries for multiple articles, either separately or all at once. You can use the platform to compare papers side by side, analyze thematic overlaps, or even merge multiple summaries into one cohesive overview. Another strong point of SciSummary is its built-in references section. You can easily view the reference list of any uploaded paper, and for each citation, there’s an “Import and Summarize” button that lets you pull that source into your reading list and generate a summary right away.

This is a fantastic way to grow your literature base organically, simply by exploring what your current papers are citing. On top of that, SciSummary’s recommendation feature suggests additional papers related to your topic, giving you a chance to discover resources you may have missed in your initial literature search.

ChatPDF

ChatPDF is another powerful AI tool that can significantly enhance how you engage with academic literature. As the name suggests, it’s built around the idea of “chatting” with your PDFs. All you need to get started is a free account and PDF copies of the papers you want to work with. Once logged in, simply drag and drop your document, and you’re ready to begin interacting with it. You can ask ChatPDF just about anything related to the content whether it’s for clarification on a term, a quick summary of a section, or deeper insight into a methodology or conclusion. It even suggests potential questions to help guide your exploration, which is especially helpful when you’re not sure where to start.

If you’re looking to analyze multiple PDFs at once, you can create a folder, drop your selected documents into it, and then chat with all of them collectively. Another feature that stands out is the ability to highlight any paragraph within your PDF and instantly access options like explain, summarize, or rewrite. This hands-on functionality makes it easy to dive deeper into specific sections and can be a great support tool when grappling with dense or technical content.

 ChatDoc

ChatDoc functions in a similar way to ChatPDF but adds several extra features that make it even more versatile and user-friendly. One of the key differences is its support for multiple file formats, you’re not limited to just PDFs. ChatDoc accepts files in PDF, DOC, DOCX, TXT, and EPUB formats, which gives you more flexibility depending on where and how you access your academic readings. You can even upload files using a URL, which is particularly useful if you’re working with content hosted online. Another key feature is its ability to read scanned documents using Optical Character Recognition (OCR), and even interact with the content of full web pages.

ChatDoc also allows you to chat with multiple documents at once. You can create a “collection,” add several files to it, and then interact with the entire collection simultaneously. This is handy if you’re comparing different papers or trying to synthesize ideas across multiple sources. On top of that, ChatDoc includes useful extras like formula recognition and image drawing support, which add another layer of interactivity particularly helpful if you’re working with mathematical or visual content. And if you’re using Chrome or Edge, there’s even a browser extension that allows you to bring ChatDoc with you as you browse. You can chat with online articles, reports, or any webpage directly blending reading and inquiry into a single seamless experience.

SciSpace

SciSpace is an AI-powered platform that brings together several academic tools in one place. It includes features like a paraphraser, AI writer, citation generator, AI detector, and, most relevant for our discussion, the ‘Chat with PDF’ functionality. This feature allows you to engage directly with your research papers making it easier to understand and analyze complex academic content.

Using SciSpace’s Chat with PDF tool is simple. Once you upload your PDF, a chat panel appears on the right-hand side where you can begin interacting with the document. SciSpace offers a curated set of sample questions you can ask right away. The Copilot tool is another helpful feature. You can highlight any sentence or paragraph in the PDF and click to get an instant explanation. If you’re looking to expand your research the platform also includes a Related Papers function that surfaces similar studies, which can be a helpful addition to your literature base.

AskYourPDF

As the name suggests, AskYourPDF is an AI-powered tool that lets you have interactive conversations with your PDF documents. The tool supports a range of file formats beyond PDFs, including TXT, PPT, PPTX, EPUB, RTF, and more giving you plenty of flexibility in how you use it. Getting started is easy, just upload your document and use the document widget to begin asking questions and exploring its contents. From the moment you upload a paper, AskYourPDF suggests questions based on its content, helping you get started with guided inquiry. Of course, you can also pose your own questions depending on what you’re looking to understand.

Scite

Scite brings a unique strength to the reading process with its browser-based extension. We find its Chrome extension particularly helpful for analyzing research articles in real time. As you browse journal websites or digital libraries, Scite adds valuable context through its Smart Citations feature. These citations don’t just list which papers cited a given study, they show how it was cited. You’ll see whether a paper was supported, contradicted, or simply mentioned in subsequent research. This immediate feedback is useful when trying to gauge a study’s credibility or relevance to your own work.

Scite also allows you to trace the citation network around a paper. With just a click, you can explore how a piece of research fits into the larger scholarly conversation, revealing trends, gaps, or methodological shifts over time. It’s especially handy when you’re building a narrative for your literature review and need to understand how different findings are connected or where they diverge.

Conclusion

Reading the literature is the intellectual foundation upon which your research study is built. It’s where you begin to form your understanding of the field, identify gaps, and develop the critical lens that will guide your inquiry. In this chapter, we explored how AI can enhance this process not by doing the reading for you, but by acting as a co-thinker that supports, extends, and deepens your engagement with academic texts.

We began with tools like NotebookLM, ChatGPT, and Claude, which share a powerful common feature: they allow you to build a custom knowledge base from the very sources you’re working with. Instead of relying on generic AI responses, you can upload your own documents and have the AI assist you within the context of your research materials. It’s a shift from working on information to working with it. These platforms offer dedicated workspaces “notebooks” or “projects” where your documents, conversations, and insights live together in a coherent, evolving research environment.

This approach allows you to ask questions, synthesize ideas, compare perspectives, and even test your own understanding all from within your own curated body of knowledge. Beyond these foundational platforms, we also explored a range of additional AI tools, each offering unique features from interactive figure explanations to audio summaries and multilingual support. But with all these tools at your disposal, the most important principle remains unchanged: AI is here to assist, not replace. It’s your analysis, your synthesis, and your critical judgment that ultimately shape your understanding and your literature review.