Copyright Notice

These slides are distributed under the Creative Commons License (CC BY-NC-SA 4.0).

​

Bingyang Wei makes these slides available for educational purposes. You are permitted to make copies and distribute these slides for non-commercial, educational purposes, provided that Bingyang Wei is cited as the source. Commercial use of these slides is not permitted.

​

For full license details, please visit https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode

Welcome

Empowering Faculty with AI for Scholarship and Teaching Workshop Series

Workshop 3

AI for Research and Grant Writing

Rethink Work: AI Automates Tasks, Not Jobs

• A job can be viewed as a set of tasks, some of which are more susceptible to automation using AI than others.
• Most jobs consist of a mix of automatable and non-automatable tasks.
• AI is not replacing entire jobs—it’s automating specific tasks within jobs.

1. Brynjolfsson, Erik, Tom Mitchell, and Daniel Rock. "What can machines learn and what does it mean for occupations and the economy?." AEA papers and proceedings. Vol. 108. 2014 Broadway, Suite 305, Nashville, TN 37203: American Economic Association, 2018.
2. Handa, Kunal, et al. "Which Economic Tasks are Performed with AI? Evidence from Millions of Claude Conversations.“ Anthropic, 2025.

General Physician

✅ High

✅ Medium

✅ High

❌ Low

❌ Low

What Does a Researcher Do Every Day?

• Finding Research Ideas
• Validating Research Ideas
• Conducting a Literature Review
• Processing and Analyzing Data
• Drafting Papers
• Reviewing Papers
• Drafting Grant Proposals
• Reviewing Grant Proposals
• Mentoring Student Researchers
• …

Finding Research Ideas

AI can help this task in three ways:

• Write a paragraph about a research problem and ask the LLM to brainstorm ideas:
• LLMs are good at suggesting interdisciplinary connections and novel angles thanks to the general knowledge they were trained on.
• Existing problems may now be solved by AI thanks to the recent breakthrough.

AI-Suitable? ✅ High

Click here to see an example

How to Determine If Your Research Problem is Suitable for AI?

• Do you have enough high-quality training data in the format of (input, output) pair?
• If so, you may frame your research problems as machine learning problems.

Research Problem: Bat Call Identification

Input

Output

(Common Pipistrelle, Echolocation Call)

(Kuhl’s Pipistrelle, Echolocation Call)

(Common Pipistrelle, Feeding Buzz)

(Common Pipistrelle, Social Call)

Input: audio recording of bat calls

Output: (species, sound classification)

… …

BTO (British Trust For Ornithology) Acoustic Pipeline

Research Problem: Tumor Tissue Detection

Input

Output

0.92

Input:

1. DNA/RNA
2. Protein
3. Phosphate
4. Lipids
5. Fatty Acids
6. Amide I
7. Lipids
8. Proteins/Lipids
9. C-H Bonds

(0.58, 0.62, 0.49, 0.21, 0.31, 0.39, 0.12, 0.09, 0.11)

0.05

(0.22, 0.28, 0.35, 0.72, 0.85, 0.69, 0.89, 0.91, 0.87)

0.52

(0.48, 0.50, 0.45, 0.48, 0.52, 0.49, 0.41, 0.39, 0.36)

0.02

(0.20, 0.24, 0.32, 0.78, 0.91, 0.74, 0.95, 0.97, 0.92)

Kothari, Ragini, et al. "Raman spectroscopy and artificial intelligence to predict the Bayesian probability of breast cancer." Scientific reports 11.1 (2021): 6482.

… …

Output: Cancerous or not

Research Problem: Translation

Input

Output

Google Translate

Output: Target language

Input: Source language

… …

Research Problem: Protein Structure Prediction

Input

Output

MVLSEGEWQLVLHVWAKVEADVAGH

MNIFEMLRIDEGLRLKIYKDTEGYYTIGIGHL

GTGGAATGGAAC

… …

Jumper, John, et al. "Highly accurate protein structure prediction with AlphaFold." nature 596.7873 (2021): 583-589.

Output: 3D structure

Input: Amino acid sequence

Research Problems As (input, output) Pair

Thinking About Data in Your Field

• Research Questions:
• What kinds of datasets exist in your field?
• What patterns or predictions could AI uncover from these datasets?
• Are there manual, time-consuming tasks in your research that AI could help automate?
• AI models are only as good as the data you feed them.
• Get inspiration from public datasets
• Kaggle
• Hugging Face

AI Implementation

• Approach 1: DIY - Learn & Implement AI Yourself
• Take AI Courses: Deep Learning Specialization, Machine Learning Specialization by Andrew Ng (requires Python knowledge)
• This approach gives you full control but requires time to learn programming and AI fundamentals.
• Approach 2: Collaborate with Computer Scientists
• Frame your research as an ML problem – Clearly define input, output, and data.
• Hire CS students – They can train and implement AI models for your research.
• Co-author with AI experts – Work with CS faculty to apply AI in your field.
• Approach 3: Leverage Your Institution’s AI Center
• Many universities have AI Centers that provide technical support, AI tools, and consulting services.
• Use pre-trained models and cloud-based AI solutions without coding.
• If no AI Center exists, advocate for AI infrastructure to support interdisciplinary research.

What Does a Researcher Do Every Day?

• Finding Research Ideas
• Validating Research Ideas
• Conducting a Literature Review
• Processing and Analyzing Data
• Drafting Papers
• Reviewing Papers
• Drafting Grant Proposals
• Reviewing Grant Proposals
• Mentoring Student Researchers
• …

Validating Research Ideas with ChatGPT

Come up with an idea

Ask AI for feedback and a draft proposal

Conduct a literature review

Or refine the idea

Idea

Research Proposal

Lit. Review

What Does a Researcher Do Every Day?

• Finding Research Ideas
• Validating Research Ideas
• Conducting a Literature Review
• Processing and Analyzing Data
• Drafting Papers
• Reviewing Papers
• Drafting Grant Proposals
• Reviewing Grant Proposals
• Mentoring Student Researchers
• …

Conducting a Literature Review

• Searching Papers: Perplexity, Elicit, Research Rabbit, scite.ai
• Knowledge Management: NotebookLM by Google
• Chatting with papers:
• Summarizing existing research
• Extracting key arguments
• Identifying research gaps
• Comparing methodologies across multiple papers

AI-Suitable? ✅ High

Click here to see an example

What Does a Researcher Do Every Day?

• Finding Research Ideas
• Validating Research Ideas
• Conducting a Literature Review
• Processing and Analyzing Data
• Drafting Papers
• Reviewing Papers
• Drafting Grant Proposals
• Reviewing Grant Proposals
• Mentoring Student Researchers
• …

Processing and Analyzing Data

• LLMs are good at various natural language processing tasks:
• Summarization & Thematic Analysis
• Sentiment Analysis & Opinion Mining
• Information Extraction
• Text Cleaning & Preprocessing
• Customizing an LLM (Covered in workshop 2)
• Retrieval-Augmented Generation (RAG)
• Model Fine-Tuning

AI-Suitable? ✅ High

What Does a Researcher Do Every Day?

• Finding Research Ideas
• Validating Research Ideas
• Conducting a Literature Review
• Processing and Analyzing Data
• Drafting Papers
• Reviewing Papers
• Drafting Grant Proposals
• Reviewing Grant Proposals
• Mentoring Student Researchers
• …

Drafting Papers

• The Snowball Approach: Writing a Paper with AI
• Start with a Plan: Ask AI to generate an outline or roadmap.
• Work Section by Section: Focus on one part at a time (introduction, methods, discussion).
• Iterate & Refine: Provide feedback, adjust content, and ensure logical flow.
• Leverage Context: AI improves when it references previously generated sections (research questions, related work, etc.).
• Tools:
• ChatGPT Canvas
• Creating visuals: Napkin
• Useful Prompts:
• ChatGPT Prompts for Academic Writing
• ChatGPT-Academic-Prompt

Drafting Papers – Caveat ⚠️

• AI-generated citations can be unreliable
• Always verify references with original sources before including them in your work.
• AI may introduce misinformation
• Cross-check key arguments, claims, and data with peer-reviewed sources.
• AI Plagiarism
• Academic Integrity & AI Detection
• Journals and universities have strict policies on AI use in writing.
• Some publishers require disclosure of AI-assisted content. Check your institution’s guidelines.
• AI-generated content can sometimes be flagged by AI detection tools like ZeroGPT, GPTZero, Grammarly, QuillBot.

Publisher’s Authorship Policy on GenAI ⚠️

• “Generative AI tools and technologies, such as ChatGPT, may not be listed as authors of an ACM published Work. The use of generative AI tools and technologies to create content is permitted but must be fully disclosed in the Work. For example, the authors could include the following statement in the Acknowledgements section of the Work: ChatGPT was utilized to generate sections of this Work, including text, tables, graphs, code, data, citations, etc.). If you are uncertain ­about the need to disclose the use of a particular tool, err on the side of caution, and include a disclosure in the acknowledgements section of the Work.” - ACM
• “The use of artificial intelligence (AI)–generated text in an article shall be disclosed in the acknowledgements section of any paper submitted to an IEEE Conference or Periodical. The sections of the paper that use AI-generated text shall have a citation to the AI system used to generate the text.” - IEEE
• “If you are using generative AI software tools to edit and improve the quality of your existing text in much the same way you would use a typing assistant like Grammarly to improve spelling, grammar, punctuation, clarity, engagement or to use a basic word processing system to correct spelling or grammar, it is not necessary to disclose such usage of these tools in your Work.” - ACM

What Does a Researcher Do Every Day?

• Finding Research Ideas
• Validating Research Ideas
• Conducting a Literature Review
• Processing and Analyzing Data
• Drafting Papers
• Reviewing Papers
• Drafting Grant Proposals
• Reviewing Grant Proposals
• Mentoring Student Researchers
• …

Reviewing Papers

• Use LLMs to mock review of your own paper
• Publishers generally prohibit the use of GenAI tools in reviewing research manuscripts due to concerns about confidentiality, integrity, and the potential for inaccurate assessments.

What Does a Researcher Do Every Day?

• Finding Research Ideas
• Validating Research Ideas
• Conducting a Literature Review
• Processing and Analyzing Data
• Drafting Papers
• Reviewing Papers
• Drafting Grant Proposals
• Reviewing Grant Proposals
• Mentoring Student Researchers
• …

Drafting Grant Proposals

• Curated AI Resources for Grant Writing
• AI for Grant Writing by a Stanford professor
• AI and Grant Preparation by Texas Tech Library
• Understanding the call for proposals
• Prompt: “Act as an experienced [EU/NL] grant advisor specializing in advising and writing [funding program] proposals. Perform an in-depth analysis of all the documents and summarize your key findings on the basic requirements to design a strong proposal for [funding program]. Structure your answer to make them easily understand. Your report should be in-depth, comprehensive, and succinct.”
• Proposers are encouraged to disclose their use of generative AI in grant applications, ensuring transparency and compliance with funding agency policies.

AI-Suitable? ✅ High

Tse-Hsiang Chen, Ryan Henderson and Ayla Kruis. "Generative AI Essentials: Introducing your digital colleague." October 4 2024, ARMA-NL conference

What Does a Researcher Do Every Day?

• Finding Research Ideas
• Validating Research Ideas
• Conducting a Literature Review
• Processing and Analyzing Data
• Drafting Papers
• Reviewing Papers
• Drafting Grant Proposals
• Reviewing Grant Proposals
• Mentoring Student Researchers
• …

Reviewing Grant Proposals ⚠️

• Use LLMs to mock review of your own proposal
• Funding Agency’s Regulations on GenAI
• NSF reviewers are prohibited from uploading any content from proposals, review information and related records to non-approved generative AI tools.
• NIH prohibits NIH scientific peer reviewers from using natural language processors, large language models, or other generative Artificial Intelligence (AI) technologies for analyzing and formulating peer review critiques for grant applications and R&D contract proposals.

1. https://www.nsf.gov/news/notice-to-the-research-community-on-ai
2. https://grants.nih.gov/grants/guide/notice-files/NOT-OD-23-149.html

What Does a Researcher Do Every Day?

• Finding Research Ideas
• Validating Research Ideas
• Conducting a Literature Review
• Processing and Analyzing Data
• Drafting Papers
• Reviewing Papers
• Drafting Grant Proposals
• Reviewing Grant Proposals
• Mentoring Student Researchers
• …

Mentoring Student Researchers:

• AI as the co-advisor
• Offering research guidance
• Helping coding task and data analysis
• Helping writing task
• Bridging the mentorship gap by:
• Answering common research questions anytime
• Providing instant feedback for research ideas
• Being patient and non-judgmental

AI-Suitable? ✅ High

How ChatGPT is transforming the postdoc experience, https://www.nature.com/articles/d41586-023-03235-8

Click here to see an example

AI as Your Co-Scientist

• Virtual Lab by Stanford
• A team of specialized LLM agents (biologist, ML specialist, critic)
• Key Tools: ESM, AlphaFold-Multimer, Rosetta for computational design
• Workflow: AI-guided iterative nanobody optimization and experimental validation
• AI Co-Scientist by Google Research
• Aid scientists in creating novel hypotheses and research plans

1. Swanson, Kyle, et al. "The virtual lab: AI agents design new SARS-CoV-2 nanobodies with experimental validation." bioRxiv (2024): 2024-11.
2. Gottweis and Natarajan, "Accelerating scientific breakthroughs with an AI co-scientist." 2025-02

Conclusion: AI as a Research Partner

• AI is a powerful tool for researchers, assisting in idea generation, literature reviews, drafting papers, and grant writing.
• AI enhances productivity, but human expertise, critical thinking, and ethical considerations remain essential.
• Effective use of AI requires clear prompts, iteration, and validation—it’s a collaboration, not automation.
• Be mindful of privacy and academic integrity—AI should be used responsibly, especially in reviewing confidential content.
• Moving forward: Experiment with AI tools in your workflow, explore AI for research efficiency, and stay informed about best practices.