A research team develops a large language model trained exclusively on a single, simple task: predicting the next word in a sentence, using a vast corpus of text from the internet. During evaluation, they discover the model is highly effective at identifying grammatically incorrect sentences, a task it was never explicitly trained to perform. Which of the following statements provides the best analysis of this outcome?

A research team wants to build a model that is highly proficient at identifying the grammatical structure of sentences. Based on the principles of how large models acquire linguistic skills, their most effective primary strategy would be to train the model from scratch exclusively on a dataset of sentences that have been manually annotated with their grammatical structures.

Explaining Un-Taught Abilities