Analyzing Iterative Prompting through the Lens of Human Cognition
A common strategy for interacting with a language model involves providing an initial instruction, reviewing the output, and then giving follow-up instructions to refine the result. Analyze why this multi-step, feedback-driven process is often more effective than attempting to formulate a single, comprehensive instruction from the outset. In your analysis, explicitly compare this interactive method to how humans typically approach learning a new skill or completing a complex project.
0
1
Tags
Ch.3 Prompting - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Analysis in Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
A user wants a language model to generate a short story. Their first attempt, "Write a story about a detective," produces a generic plot. The user then provides a second instruction, "Rewrite that story, but make the detective a retired librarian and set it in a small coastal town." The model's revised output is much more unique and engaging. This step-by-step refinement process is effective primarily because it:
Advising a Frustrated User
Analyzing Iterative Prompting through the Lens of Human Cognition