This is a running list of potential uses for LLMs as instruments in scientific research workflows. All of these have been not been done as of yet, but many have and others have been preposed.
Ideation
- Brainstorming
- Offering counter arguments
- Evaluating ideas
Background Research
- Summarizing articles
- Translating articles
- Explaining concepts
- Connecting article concepts
- Creating mindmaps
Coding
- Writing code
- Optimizing code
- Explaining code
- Translating code between languages
- Debugging code
Data Preparation and Analysis
- Extracting data from text
- Reformatting data
- Classifying and scoring text
- Extracting sentiment
- Simulating human subjects
- Fuzzy joining of data sets
- Create synthetic/dummy data
- Setting up models (in LaTeX)
- Writing data dictionaries
- Qualitative coding of data
- Putting data into Pandas dataframe
Writing
- Editing for grammar, vocabulary, clarity
- Generator match catchy titles
- Synthesizing wordy text
- Formatting references
- Moving to LaTeX or Markdown
Based in part on: https://www.nber.org/papers/w30957