How to use AI as a social scientist

Introduction

This blog post focuses on using AI to enhance data analysis, streamline workflow management, facilitate efficient research synthesis, and optimize report writing. We will explore specific AI tools, techniques, and applications tailored to social scientists in these domains.

Efficient research synthesis

Efficient research synthesis refers to the process of systematically reviewing and integrating existing research findings to generate comprehensive knowledge on a particular topic. The following tools are helpful in this process.

1. Consensus: Consensus is an AI-powered tool for systematic literature review and synthesis. You can use it to determine whether scientists have a consensus on a specific research question.
2. Humata: Humata is an AI platform designed to aid researchers in literature review and synthesis. It uses natural language processing and machine learning algorithms to automate article screening, data extraction, and result synthesis.
3. Research Rabbit: Research Rabbit is an AI-powered tool that facilitates literature searches and review. It uses advanced algorithms to identify relevant studies, extract critical information, and provide summarized insights, helping researchers streamline the research synthesis process.
4. Elicit: Elicit is an online platform that assists researchers in systematically reviewing and synthesizing research evidence. It offers features such as automated search, screening, and data extraction.
5. Litmaps: Litmaps is an AI-powered literature review tool that visualizes research literature. It helps researchers explore and map the landscape of existing studies, identify key themes, and discover connections and gaps in the literature.
6. Google Scholar: Google Scholar is a widely used academic search engine that provides access to a vast collection of scholarly articles. It allows researchers to search for relevant studies, explore citation networks, and access full-text articles.

Data analysis

AI-powered software can analyse large amounts of data to identify patterns and provide insights that might be missed otherwise. This kind of data analysis can help knowledge workers make informed decisions quickly and effectively. Powerful data analysis tools like R and SPSS have become more accessible due to AI being able to assist in writing code and fixing errors. The list below just partly covers all of the programming languages and syntax that Chat-GPT can write. It is not only able to write code from simple prompts but also able to make code more efficient, find errors and explain exactly what is being executed by running the code. This is of particular importance for researchers where every step in the research process has to be carefully described in plain language.

1. R:
   • Syntax: R uses a combination of functions, operators, and symbols. It follows a syntax similar to other programming languages, with functions often taking the form of function_name(arguments).
2. Python:
   • Syntax: Python uses a straightforward and readable syntax. It relies on indentation and colons to define code blocks. Functions are typically defined as def function_name(arguments): followed by an indented code block.
3. SQL (Structured Query Language):
   • Syntax: SQL is specifically designed for querying and manipulating databases. Its syntax includes keywords such as SELECT, FROM, WHERE, and JOIN for data retrieval, filtering, and joining operations.
4. SAS (Statistical Analysis System):
   • Syntax: SAS has its own programming language, which uses a combination of data step programming (for data manipulation) and proc step programming (for data analysis). Statements in SAS often start with keywords like DATA, PROC, and RUN.
5. SPSS (Statistical Package for the Social Sciences):
   • Syntax: SPSS utilizes a command syntax language for data manipulation and statistical analysis. It involves writing commands in a sequential manner, with keywords like COMPUTE, FREQUENCIES, and REGRESSION.
6. MATLAB:
   • Syntax: MATLAB uses a syntax similar to traditional programming languages. It employs functions, operators, and control flow statements. Functions are called using the format function_name(arguments).
7. Julia:
   • Syntax: Julia has a syntax that is similar to other high-level programming languages. It features functions, loops, conditionals, and mathematical operations, allowing for efficient numerical computations.
8. Stata:
   • Syntax: Stata uses its own command-driven language. It employs keywords such as generate, drop, and regress to perform data manipulation and statistical analysis.

Workflow

AI-based tools can also assist with managing emails and calendars, which can be a time-consuming task. With the help of AI algorithms, emails and messages can be prioritized based on their importance, and scheduling meetings can be automated. This can help knowledge workers save time and focus on other important tasks. When using these tools it is important to write good prompts. A good prompt consists of the following three elements:

1. Define the role: Specify the role you want the AI to take on, aligning it with a specific persona or function. For example, you can instruct the AI to act as a blogger, news article writer, academic researcher, or personal assistant. Defining the role helps the AI model understand the context and generate content that is suitable for that particular role.
2. Specify the task: Clearly communicate the task you want the AI to perform. It could include composing an email response, drafting a meeting invitation, summarizing a calendar event, or generating a blog post. Providing a straightforward task allows the AI to focus on generating content that fulfils that specific objective.
3. Define the desired format: Specify the format in which you would like the output to be presented. For instance, you can request the AI to generate content in a specific length (e.g., short paragraph, concise summary, or detailed report), provide information in a table or list format, or structure the content with appropriate headers and subheadings. Clearly defining the format ensures that the AI-generated content meets your expectations and is suitable for its intended purpose.

The tools used

1. Gmail: Gmail, a widely used email service, offers AI-based features that can help manage emails efficiently. With features like Smart Reply and Priority Inbox, AI algorithms prioritize emails based on their importance, saving time by automatically sorting and suggesting responses.
2. Notion (AI): Notion is a productivity tool that utilizes AI capabilities to enhance organization and collaboration. With AI-powered features like automated task management, content suggestions, and data organization, Notion streamlines workflow and helps social scientists stay organized and productive.
3. chatGPT: chatGPT, an AI language model, can be utilized as a virtual assistant for social scientists. It can assist in various tasks such as answering questions, providing research insights, and offering suggestions for data analysis or literature review. chatGPT’s conversational capabilities make it a valuable tool for researchers seeking quick information or guidance.

Report and article writing

In addition, AI can assist in content creation, such as writing reports and articles. AI-powered writing assistants can help in generating ideas, improving grammar and spelling, and even generating entire articles. It is crucial to approach AI-generated content with a responsible mindset. It is best to treat the AI as a “somewhat decent intern” While AI can be a helpful assistant, it is important to remember that, as the researcher, you bear full responsibility for the output produced by the AI. Therefore, it is essential to exercise caution and critical judgment when utilizing AI-generated content. The following tools are helpful when writing scientific articles.

1. Grammarly: Grammarly is an AI-based writing assistant that helps improve grammar, spelling, and style. It can be helpful for composing emails, writing reports, or refining articles. Grammarly provides real-time suggestions and corrections, enhancing the clarity and professionalism of your written content.
2. Copyscape: Copyscape is an AI tool that checks for plagiarism by comparing your written content against a vast database of online sources. It ensures that your work is original and avoids unintentional plagiarism when referencing or citing external materials.
3. Mendeley: Mendeley is a reference management tool that uses AI algorithms to organize and annotate research papers. It helps researchers discover relevant articles, create citation libraries, and collaborate with others in their field. Mendeley’s AI-powered recommendations feature suggests related papers based on your interests and reading history.
4. Zotero: Zotero is another AI-driven reference management tool that enables researchers to collect, organize, and cite sources. It assists in generating bibliographies, capturing web content, and collaborating with team members. Zotero’s AI capabilities aid in extracting metadata and streamlining the citation process.
5. RefWorks: RefWorks is a web-based reference management tool that simplifies organizing and citing sources. It allows researchers to create personal libraries, generate citations in various styles, and collaborate with colleagues. RefWorks’ AI-powered features assist in automatically extracting metadata and creating bibliographies.