Short Answer

Limitations of Supervised Fine-Tuning for Value Alignment

A team is trying to make a large language model 'ethically considerate'. Their approach is to collect a million examples of ethically sound statements and train the model to mimic them perfectly. Explain the primary limitation of this 'data-fitting' strategy for instilling a general understanding of ethical considerations in the model.

0

1

Updated 2025-10-06

Contributors are:

Who are from:

Tags

Ch.4 Alignment - 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