Question: Write a concise analytical response connecting training-set size, human insight, hand-engineered knowledge, and possible system performance.

Sample answer: When the training set is very small, an end-to-end system may not acquire enough knowledge from the available examples. A hand-engineered pipeline can incorporate knowledge derived from human insight. Because the end-to-end system lacks this hand-engineered knowledge, it might perform worse than the pipeline in the small-data setting. Thus, limited training data increases the value of hand-engineered components as a source of algorithmic knowledge.

Key points:

• A very small training set provides limited learned knowledge.
• Human insight supplies much of the needed algorithmic knowledge.
• Hand-engineered components encode that human insight.
• An end-to-end system lacking this knowledge might underperform a hand-engineered pipeline.

Rubric: A strong response accurately explains the small-data condition, identifies human insight as the source of much of the algorithm's knowledge, connects that insight to hand-engineered components, and states that the end-to-end system might—not necessarily will—perform worse.