A company is considering modernizing its NLP foundation. Across diverse natural language processing (NLP) tasks such as translation, summarization, and question answering, which characteristic MOST likely gives generative AI (a foundation model) an advantage over traditional task-specific models?
Choose the characteristic that gives generative AI an advantage in NLP.
A single model adapts to context and can be repurposed for diverse tasks
Correct. Because a foundation model is trained on broad data, a single model can adapt to and be repurposed for diverse NLP (natural language processing, the technology that lets computers handle human language) tasks. This reduces the need to build a separate model for each task.
Output is deterministic and fully reproducible
Generative AI is inherently non-deterministic, and the same input can produce different outputs.
Determinism and reproducibility are actually a weak side of generative AI, so describing them as an advantage is incorrect.
The internal reasoning of the model can be fully explained
Large foundation models are internally complex, so explaining their reasoning is actually difficult (they are opaque).
Explainability is a challenge, not an advantage, so this description is incorrect.
Events that occur after training are reflected automatically
A model's knowledge is frozen at training time, and new events are not reflected without a mechanism such as RAG.
Automatic reflection is not a property of a foundation model on its own, so this is incorrect.
Remember the correct answer: context adaptation and transferability.
・Because a foundation model is trained on broad data, a single model can handle diverse NLP (natural language processing, the technology that lets computers handle human language) tasks such as translation, summarization, and question answering by adapting to context.
・This reduces the need to build a separate model for each task.
'A dedicated model is required per task,' 'context is ignored,' and 'less data means higher accuracy' are all wrong and are the opposite of why generative AI has an advantage in NLP.