NCA-GENL Question 75
Select 3A retail company wants to improve its decision-making process for inventory management using a generative AI large language model (LLM). Which of the following approaches would best support accurate and efficient decision-making?
- A
Use the LLM to predict future inventory needs by analyzing historical sales data and trends.
- B
Generate summaries of customer feedback using the LLM to identify patterns in product preferences.
- C
Rely solely on the LLM's generated text output without validating it against real-world inventory data.
- D
Incorporate the LLM's suggestions into a larger decision pipeline that includes human oversight and traditional predictive models.
- E
Train the LLM on raw inventory data without preprocessing or cleaning to ensure faster deployment.
Show answer and explanation
Correct answers: A, B, D
Explanation
Generative AI LLMs can support decision-making by analyzing historical data, summarizing feedback, and identifying trends. However, they are most effective when their outputs are validated and integrated into a robust decision-making pipeline that includes human oversight and traditional techniques. This ensures high reliability and accuracy in the inventory management process.
- A. Correct.
Using the LLM to predict future inventory needs based on historical data is a valid use case as it can identify trends and patterns to support decision-making processes.
- B. Correct.
Generating summaries of customer feedback can help reveal product preferences, which can be critical for inventory planning and improving customer satisfaction.
- C. Incorrect.
Relying solely on the LLM's output without validation is risky, as LLMs can produce errors or hallucinations, making it an unreliable approach for decision-making.
- D. Correct.
Incorporating the LLM's insights into a larger decision pipeline ensures that human expertise and traditional models validate and enhance the AI's outputs, leading to better decision-making.
- E. Incorrect.
Training the LLM on raw data without preprocessing introduces noise and inconsistencies, which can degrade the quality of the generated insights, thereby making it unsuitable for accurate decision-making.