Data Analysis and Visualization with Generative AI

Generative AI tools can be very helpful assisting in writing code, conducting data analysis, and creating data visualizations. AI algorithms are quite effective at tasks related to correcting and enhancing code. As with any AI tool, the output of any AI data tool should be vetted for accuracy.  

Many experienced software developers and data scientists use AI tools in their work. However, using an AI tool to write code from scratch can be dangerous, especially for beginners. Beginners should use caution when using AI tools to assist in coding and working with data, as beginners face:

• Lack of understanding of coding syntax

• Inadequate experience and language to craft useful prompts

• Not enough proficiency to vet the output of code generated by AI

Rather than writing code or analyzing data from scratch, AI tools are most effective when used to generate ideas, enhance existing code, or save time. Researchers are responsible for creating a research framework, deciding on analysis methods, selecting visualizations, and crafting code that executes the planned analysis and visualization. 

Generative AI can be a helpful assistant, but you must stay in charge of your analysis, visualization, and code. You, as the researcher, are responsible for reviewing, vetting, and validating any AI-generated content. 

What CAN’T AI do to assist in writing code for data analysis and visualization?

• Anything that is prohibited by your department and/or course syllabus

• Access current information (for example, ChatGPT's knowledge cutoffs are noted in the product documentation)

• Create “ready-made” code, tailored to your exact computer, files, variables, etc.

• Decide which visualizations, tables, etc. are most compelling and support your argument

• Determine which statistical test is most appropriate for your analysis/data

How can I use AI to assist in writing code?

• Help you understand error messages

• Identify and correct pesky errors in code

• Generate ideas for new packages or libraries to use

• Demonstrate uses of functions with example variables (can be helpful for packages/libraries with confusing documentation)

Questions about using AI to write code, or assist in data analysis or visualization? Email digitalscholarship@georgetown.edu.

Questions about using Machine Learning, AI, and Natural Language Processing tools for text analysis? Check out the Text Mining LibGuide.