I am struggling with Cognitive Outsourcing

Okay, to start off, just a little bit of my background. I am a master’s student at UCSD pursuing a degree in computer science. This hit me when, during this summer, I tried to outsource my creative thinking and analysis for a research project that I am working on to an LLM, and the results (to no one’s surprise) were horrible in that I had no clue of what I was talking about and made a complete fool of myself while presenting “my work” to my supervisor. The excuse I kept giving myself was that everyone uses Claude nowadays, and if I don’t integrate it into my workflows, I will be left behind. But instead of that, what I was actually doing was offloading all my tasks and responsibilities as a “human” onto a glorified autocomplete.

This realization has been deeply troubling to me and something that I am still thinking about as I write my thoughts here. Reading a recent report by MIT reveals a “fragmented and conflicted sense of authorship” for a control group allowed to use LLMs to write essays for the study vs. the “brain-only” group, which used no external tools. [1] My point being, as a researcher, am I truly contributing anything meaningful if all I did is just prompt an LLM and claim its output as my own and only gain superficial understanding of a concept?

Work in Progress

References

• Nataliya K, et al. “Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task.” arxiv:2506.08872