How to use ChatGPT for Research - Literature Review, Research Paper Writing & Publication

TL;DR

ChatGPT can be used across the research lifecycle: literature review, topic selection, research-gap identification, paper synthesis, experimental planning, data analysis, writing support, figure ideation, and journal shortlisting.

Briefing Cornell Notes

Briefing

ChatGPT is positioned as an end-to-end “research assistant” that can support nearly every stage of academic work—from finding and narrowing a topic to analyzing data, drafting a paper, creating figures, and even identifying where to submit. The practical pitch is that researchers can use it without paying for a Pro plan, and can tailor prompts to get outputs that are grounded in scholarly literature rather than generic web content.

The workflow begins with literature review. A researcher can ask for high-impact open access review papers on a specific subject (for example, superhydrophobic coatings using nanoparticles) and explicitly request free PDFs so the sources are accessible. From there, ChatGPT can help narrow a research direction by recommending promising materials and surfacing research gaps—again with a constraint to rely only on peer-reviewed articles. The transcript emphasizes an important prompt detail: adding “refer to only peerreviewed articles” to reduce the risk of answers being built from random blogs and to keep the results evidence-based.

Once candidate papers and a topic are identified, the next step is synthesis. The transcript describes uploading multiple papers and asking for comparative analysis in a structured table—comparing methodology, efficiency, and durability, and highlighting which paper offers the most promising results. This is framed as especially useful for systematic review writing, where cross-paper comparison matters more than isolated summaries. A limitation is noted for free users: there’s a cap on how many papers can be uploaded per day.

After topic selection and paper understanding, ChatGPT can support experimental planning and instrumentation. For sample characterization using an atomic force microscope (AFM), the transcript describes asking for step-by-step sample preparation and scanning guidance tailored to the researcher’s material system, along with tips for preparing superhydrophobic coatings. For data analysis, it can generate visualizations such as 3D surface plots and compute roughness values from shared measurements.

Writing and publication tasks come next. ChatGPT can generate outlines, improve language and grammar, paraphrase uncertain sections, and help make drafts more academic—while the transcript cautions against outsourcing entire paragraphs or whole papers to AI. It also includes a privacy workflow: before sharing sensitive data, users should go to settings and data control and switch off “improve the model for everyone” to prevent training use or exposure.

Finally, the transcript extends ChatGPT’s role into research communication and submission strategy. It can generate graphical abstracts from an uploaded abstract and can refine rough sketches into more polished illustrations. For journal selection, it can recommend Scopus indexed peer-reviewed journals based on the abstract. Taken together, the guidance presents ChatGPT as a flexible toolchain for research planning, execution support, manuscript development, and publication logistics—so long as prompts are specific and outputs are checked against credible sources.

Cornell Notes

ChatGPT is presented as a practical “research assistant” that can help across the full academic workflow: literature review, topic narrowing, research-gap identification, paper synthesis, experimental planning, data analysis, manuscript drafting, figure ideation, and journal shortlisting. The transcript highlights prompt tactics that keep answers grounded in scholarship—such as requesting open access PDFs and adding constraints to use only peer-reviewed articles. It also describes uploading papers for comparative, table-based analysis of methodology, efficiency, and durability, and using ChatGPT to generate AFM-related sample prep and scanning guidance. For writing, it recommends using AI for outlines and language improvement while avoiding full-paragraph generation, and it advises turning off “improve the model for everyone” before sharing sensitive data.

How can a researcher use ChatGPT to start a literature review while ensuring sources are accessible and credible?

Use a prompt that specifies both the topic and access needs—e.g., asking for “high impact open access review papers” and requesting free PDFs. Then, when narrowing the research direction, add constraints like “refer to only peerreviewed articles” to steer outputs toward journal literature rather than random online material. The transcript’s example focuses on superhydrophobic coatings using nanoparticles and emphasizes checking that the listed sources are journal articles.

What prompt strategy helps ChatGPT identify a research gap without drifting into non-scholarly sources?

Include an explicit restriction to scholarly literature, such as “Please refer to only peerreviewed articles for this search.” In the transcript’s workflow, this constraint is used while asking for promising nanomaterials and for research gaps tied to those materials, with the expectation that the output will cite peer-reviewed journal articles.

How does uploading multiple papers change what ChatGPT can do for synthesis?

Uploading papers enables comparative analysis across studies. The transcript describes asking for a table comparing methodology, efficiency, and durability, and then identifying which paper provides the most promising results (e.g., “paper number four” in the example). This approach is framed as useful for systematic reviews, though free plans may limit the number of papers that can be uploaded per day.

In what way can ChatGPT support hands-on experimental work and instrumentation like AFM?

By asking for step-by-step, sample-specific guidance. The transcript’s example involves using an atomic force microscope (AFM) to measure roughness and requesting instructions for sample preparation and how to scan the sample, along with tips for preparing superhydrophobic coatings. The output is treated as a planning aid for experimentation rather than a substitute for lab protocols.

What safeguards are recommended when using ChatGPT for writing and sensitive research data?

For writing, the transcript recommends using AI as support—outlines, grammar, paraphrasing, and academic tone—while avoiding generating entire paragraphs or whole sections wholesale. For privacy, it advises checking settings and data control and switching off “improve the model for everyone” before sharing sensitive data, to reduce the chance of training use or leakage.

How can ChatGPT help with research communication and publication logistics after the manuscript draft exists?

It can generate graphical abstracts from an uploaded abstract and can also turn rough sketches (e.g., a nanoparticle-coated surface concept) into more polished illustrations. For publication planning, it can suggest Scopus indexed peer-reviewed journals based on the abstract, producing a shortlist aligned with the research area.

Review Questions

What specific prompt constraints in the transcript are meant to keep ChatGPT outputs grounded in peer-reviewed literature, and why do they matter?
How would you design a comparative prompt for multiple uploaded papers to extract methodology, efficiency, and durability differences in a table?
What privacy setting does the transcript recommend changing before sharing sensitive data, and what writing behavior does it caution against?

Key Points

1
ChatGPT can be used across the research lifecycle: literature review, topic selection, research-gap identification, paper synthesis, experimental planning, data analysis, writing support, figure ideation, and journal shortlisting.
2
Request open access and free PDFs when starting a literature review to ensure sources are actually retrievable.
3
Add “refer to only peerreviewed articles” to reduce reliance on non-scholarly web content when identifying materials and research gaps.
4
Upload multiple papers to get structured, table-based comparisons of methodology, efficiency, and durability, then identify the most promising results.
5
Use ChatGPT for lab planning support—such as AFM sample preparation and scanning guidance—while treating it as assistance rather than a replacement for established protocols.
6
For manuscript work, rely on outlines and language improvement, but avoid generating entire paragraphs or whole sections directly.
7
Before sharing sensitive data, disable “improve the model for everyone” in settings/data control to limit training or exposure risks.

Highlights

A single prompt can steer ChatGPT toward evidence-based answers by explicitly requiring peer-reviewed sources and open access PDFs.

Uploading multiple papers enables side-by-side comparison in a table format, including methodology, efficiency, and durability—useful for systematic reviews.

ChatGPT can generate AFM-related step-by-step sample preparation and scanning guidance tailored to a researcher’s material system.

The transcript recommends turning off “improve the model for everyone” before sharing sensitive data and cautions against full-paragraph AI drafting.

Graphical abstracts and journal shortlists can be generated from an abstract, including Scopus indexed peer-reviewed journal suggestions.

Topics

ChatGPT Research Workflow
Literature Review Prompts
AFM Roughness Measurement
Research Paper Drafting
Graphical Abstracts

Mentioned

Niha Graval
AFM
Scopus