NCA-GENL Question 77
Select 3A retail company wants to implement a generative AI-powered recommendation system to support decision-making for product restocking. Which of the following approaches should the company use to ensure the model provides actionable insights while maintaining efficiency?
- A
Fine-tune a pre-trained large language model (LLM) using historical sales data and product inventory records.
- B
Use a generative AI model to create synthetic sales data for training other machine learning models.
- C
Incorporate prompt engineering to query the generative AI model for specific restocking recommendations based on real-time inventory data.
- D
Leverage a rules-based system alongside the generative AI model to validate its recommendations before making decisions.
- E
Train a generative AI model from scratch using only the company’s proprietary data.
Show answer and explanation
Correct answers: A, C, D
Explanation
To support decision-making for product restocking using generative AI, the company should fine-tune a pre-trained model with relevant data, use prompt engineering for effective querying, and validate outputs using a rules-based system. These approaches balance efficiency, accuracy, and practicality while minimizing resource usage. Training a model from scratch or relying on synthetic data are suboptimal solutions in this context.
- A. Correct.
Fine-tuning a pre-trained LLM with relevant historical sales and inventory data ensures the model is specialized for the company’s needs and can provide insights specific to its operations. This is a recommended approach.
- B. Incorrect.
Using generative AI to create synthetic sales data might not yield meaningful or accurate insights for decision-making, as it risks introducing biases or inaccuracies into the training data.
- C. Correct.
Prompt engineering helps query the model effectively to generate specific, actionable insights, such as recommendations for restocking. This is an essential approach to making generative AI useful in real-world scenarios.
- D. Correct.
Combining a rules-based system with generative AI ensures that any recommendations are validated and aligned with business policies or constraints, making the decision-making process more robust.
- E. Incorrect.
Training a model from scratch using only proprietary data is resource-intensive and unnecessary when pre-trained models can be fine-tuned for the specific problem, making this approach inefficient and impractical.