Question: What type of training data is used when teaching large language models to identify patterns between words, concepts, and phrases?
- Real-time human conversations
- Thousands of photos and illustrations
- Interactions with physical objects
- Vast amounts of text
Explanation
Large language models learn relationships in language by processing text-based patterns across words, phrases, and concepts. This training helps the model predict likely language sequences and generate coherent responses. Text data is the relevant input because the task focuses on language understanding and generation. Images, physical interactions, and live conversations are not the core training source for this type of model.
Why the other options are incorrect
Real-time human conversations is incorrect because live interaction is not the standard training source for identifying broad language patterns.
Thousands of photos and illustrations is incorrect because visual data supports image-related models, not core language-pattern training.
Interactions with physical objects is incorrect because physical experience is not required for training large language models.
Source for verification
https://cloud.google.com/learn/what-is-artificial-intelligence
The answer(s) to the question is highlighted in the BOLD text above. You can also find more questions and answers related to the exams on the "Generative AI for Educators" page.