| Steps of the research process | Recommendation | Grade | Related posts |
|---|---|---|---|
| Choosing the right GenAI model | Model selection decisions should reflect the relevant tasks as well as the respective importance of speed, reasoning capabilities or price | - | model comparison |
| Brainstorming | Searching for basic ideas on the given topic | A | brainstorming |
| Searching for literature | Use Google Scholar/keyword search in Publish or Perish/ResearchRabbit/SciSpace etc | C | searching for literature |
| Write-up of the literature review | Writing up the literature review with pre-selected material | D | literature review |
| Formulating theory: Research questions and hypotheses | Can use prompt engineering to extract research questions and hypotheses from the articles | C | research questions |
| Selecting the methodology | GenAI can provide suggestions but researcher must validate outputs | B | methodology |
| Data collection | Not advised unless done with the strictest of supervision, but deep research capabilities are expanding rapidly | C | data collection |
| Data extraction | Useful for parsing and applying OCR to extract information from files | B | OCR |
| Data pre-processing | Useful for data cleaning tasks | B | data pre-processing data cleaning |
| Data analysis | Useful for certain NLP tasks, and data analysis (with supervision). Do not use in methods where you lack sufficient knowledge to validate results | C | data analysis classification Named Entity Recognition clustering |
| Data visualisation | Can summarise complex patterns, highlight key relationships and trends. Validate all visual interpretations with domain expertise. | C | data visualisation |
| Evaluating the results | Use if you have domain knowledge and can validate the output | E | results evaluation |
| Writing and formatting the article | Grammar, style corrections, and standardising citations. Machine translation | A | writing and formatting citation |
| Publication | Great for generating abstracts, less so for journal selection | B | abstract journal selection |
| Peer review | Useful for generating basic ideas and for addressing logical errors | B | peer review |
| Dissemination | Useful for sharing research in a social media-friendly format | B | news generation social media |
| Grant writing | Useful for extracting key information from calls, drafting initial versions of proposals, and checking alignment with funder expectations. | B | grant writing |
