I am not familiar with this tool, but appreciate you sharing it! I love the idea of using it to help you generate a task list when you are feeling stuck, but totally agree with your points about the task sizes not being remotely comparable. Also, there are whole parts of my research where the actual work is quite intuitive and it doesn’t lend itself well to this kind of structure (developing a theoretical framework, yes; but also parts of revising my own 1st/2nd drafts where the ‘task’ is probably just “Read it from the top and make modifications as I go” but the reality is deeply engaging with the argument and structure of the piece as a whole).
I’ve played with ChatGPT as a coding partner when doing statistical analysis in R and it was largely helpful with basic tasks (which for me is good because stats isn’t my area of expertise even though I use them in some of my research), but the code it generated for me to do complex tasks was not always useable. I’ve heard from more quantitative colleagues that this pattern holds true, and so while it can speed up some things A LOT, it can also force you to lose a day or two to checking its code to see if what it is trying to tell you do is even possible.
Thanks for creating space for these conversations, though! AI is here and I think we’ve got to figure out how to make it work ethically / effectively and model that for students.
I am not familiar with this tool, but appreciate you sharing it! I love the idea of using it to help you generate a task list when you are feeling stuck, but totally agree with your points about the task sizes not being remotely comparable. Also, there are whole parts of my research where the actual work is quite intuitive and it doesn’t lend itself well to this kind of structure (developing a theoretical framework, yes; but also parts of revising my own 1st/2nd drafts where the ‘task’ is probably just “Read it from the top and make modifications as I go” but the reality is deeply engaging with the argument and structure of the piece as a whole).
I’ve played with ChatGPT as a coding partner when doing statistical analysis in R and it was largely helpful with basic tasks (which for me is good because stats isn’t my area of expertise even though I use them in some of my research), but the code it generated for me to do complex tasks was not always useable. I’ve heard from more quantitative colleagues that this pattern holds true, and so while it can speed up some things A LOT, it can also force you to lose a day or two to checking its code to see if what it is trying to tell you do is even possible.
Thanks for creating space for these conversations, though! AI is here and I think we’ve got to figure out how to make it work ethically / effectively and model that for students.