Two teams are tasked with enabling a general-purpose Large Language Model to write accurate code in a newly developed, obscure programming language.

• Team Alpha focuses exclusively on prompt engineering. They design highly detailed prompts that include the complete language syntax, style guides, and several examples of correct code, hoping the model can learn 'on the fly' from the context provided.
• Team Beta ignores prompt engineering initially. Instead, they collect a large corpus of code written in the new language and use it to continue training the base model.

Evaluate the two strategies. Which team's approach is more likely to succeed in the long run, and why? Your explanation should focus on the relationship between a model's training data and its capabilities.