a year ago
- Students often use ChatGPT for assignments, producing verbose, bullet-point-heavy answers that lack originality.
- Reasons for using language models include perceived unimportance of the task, belief in superior model output, and lack of personal investment.
- Writing should communicate original thoughts; using models is worse than plagiarism as it conveys no human thought.
- Model-generated content often obscures meaning, adds fluff, and lacks the depth of human understanding.
- An experiment with Google Gemini showed model output to be boring, verbose, and lacking substance compared to the original prompt.