I always tell this to my research assistants: if you use AI to summarize and write for you, you don't need to be my assistant. You don't need to do research at all.
Research should be selfish: *you* (not a machine) learn how to read, write, innovate, and dream. Otherwise, what is the point?
Research should be selfish: *you* (not a machine) learn how to read, write, innovate, and dream. Otherwise, what is the point?
Comments
I am talking to someone well-established who co-authored 20+ pieces in 2024. I *barely* co-authored 3 last year.
My desire to teach learning a certain way puts my students at a numerical disadvantage. They won't get the High Score in research output.
For this reason, I'm rather conservative - if LLMs provide long-term benefit, they will still do so at a later time. If LLMs hamper students' ability to learn, I'd do them a lifelong disservice by encouraging their use.
Reading closely is a luxury I *at the very least* want students to experience before they run off and use LLMs for everything in their life.
For doing my due diligence to teach them what I know from my own experience, I am obligated to tell them that close reading is important for building foundational skills.
I lean largely on thoughts and ideas from core books and papers I read closely, some over a decade ago, but rarely ones I skimmed because I was doing research tasks.
But also, I was a philosophy student. So efficiency and output and task completion weren't motivations for my learning and writing.
This is easy if you simply view it as a task that needs to be completed and use AI. But if you do, you don't deeply learn!
Some struggle is essential to good. Don't let everything become machine management.
So happy to hear you'll take this into a classroom.
And those screenshots are from this doc I give my students: (feel free to copy, reuse, rehash if anything is useful!)
If the goal is producing papers, then LLMs are an easy choice. If the goal is deeply learning a subject, then LLMs might not be the best tool (*especially* for someone new).