AI in STAT 9610

AI tools like ChatGPT will accelerate statistics and data science in the coming years. On the other hand, these tools often confidently give wrong answers, and must be used with caution. For more context, students are encouraged to read this overview of AI tools as of summer 2024, written by Ethan Mollick of the Wharton Management Department. In STAT 9610, students are encouraged to use these tools to enhance their learning and productivity. As discussed in the Syllabus, students may use any AI tools to complete their assignments and assessments, as long as they disclose which tools they used and how they used them.

Tools

• ChatGPT. ChatGPT Enterprise will be available to all STAT 9610 students this fall; see this page for instructions and resources.

• GitHub Copilot. This AI-based autocomplete tool is free for students (see these instructions for getting access). It can be used within RStudio (for R) or within Visual Studio Code (for LaTeX). You can prompt GitHub Copilot with code comments.

Relevant use-cases

• Programming. Programming is one of AI’s current strengths. This makes it less important for programmers to memorize or look up programming syntax. I recommend drafting code chunks using GitHub Copilot, because it is conveniently integrated into RStudio. If this is insufficient, especially for longer or more complex code chunks, ask ChatGPT to write the code for you and paste it into RStudio. If you have errors, then paste both your code and your error messages into ChatGPT. If this doesn’t help, then Google your error message. Sometimes, websites like Stackoverflow still have insights that ChatGPT does not have.

• Learning statistics. ChatGPT knows about a lot of statistical concepts, and can explain them or answer questions about them.

• Editing writing. If you give ChatGPT some of your written work, it can suggest how you can make it clearer or more fluent. This is particularly useful for non-native English speakers.

• Mathematical derivations. Currently, ChatGPT is not very good at mathematical derivations.

Disclaimer

Students are responsible for the quality of the work they submit for the course. If an AI tool gives the wrong answer, it is the student’s responsibility to correct it. If an AI tool gives the right answer, it is the student’s responsibility to understand why it is right. In short, students will be expected to use AI tools to help them learn, but not to replace their learning.