Meridian's predictive maintenance model needs both the structured numeric telemetry from engine sensors and the semi-structured free-text notes mechanics type into work orders describing what they observed. What does this combination best illustrate about data for AI at Meridian?

A recurring reality in Data for AI is that meaningful AI use cases draw on a mix of data types — structured telemetry, semi-structured logs, unstructured text or images — each requiring its own preparation approach (e.g., normalization for numeric sensor data, tokenization or NLP feature extraction for mechanic notes) before being combined into a unified feature set. The claim that AI initiatives can rely on structured data alone and that free-text mechanic notes add no value wrongly dismisses the notes, which often contain early qualitative signals a sensor alone won't capture. The claim that free-text notes must be discarded because natural-language data cannot be used in an AI pipeline reflects the same 'unstructured data is unusable' misconception addressed in Q143 — text is processable with the right technique, not disqualifying. The claim that structured and unstructured data must be modeled in two entirely separate systems that never interact is an unnecessary and impractical constraint; the whole point of a mature data pipeline is to bring heterogeneous data together into one model-ready dataset, not silo it permanently. For the exam, recognize that 'data variety' isn't just a Big-Data buzzword — it directly shapes what data-preparation work (Task 4 of this domain) a PM must plan and staff for.