1.2.1. Input Modality Determines Workload Category

Different data types require fundamentally different mathematical representations. Images are grids of pixels, text is sequences of tokens, and audio is waveforms. The algorithms that process pixel grids cannot process token sequences—this is why AI splits into distinct workload categories.

The following decision tree helps you classify any exam scenario:

⚠️ Critical Exam Trap: OCR (Optical Character Recognition) is a Computer Vision workload, NOT an NLP workload! Even though the output is text, the INPUT is an image. Always classify by input type.