AIF-C01 Question 97
Select 3A company is planning to use an AWS generative AI service to create a custom language model for their chatbot. They want to minimize costs while ensuring the solution is available globally and responsive to their customers in multiple regions. Which considerations should they prioritize when designing this solution?
- A
Choose token-based pricing to control costs based on usage.
- B
Deploy the model in multiple AWS Regions to reduce latency for global users.
- C
Use a pre-trained foundation model instead of training a custom model.
- D
Provision high throughput for the model to ensure responsiveness during peak demand.
- E
Enable redundancy by deploying the model across Availability Zones in a single region.
Show answer and explanation
Correct answers: A, B, D
Explanation
To meet the company's requirements of minimizing costs, ensuring global availability, and maintaining responsiveness, the solution should focus on token-based pricing to manage costs, deploying the model in multiple regions to reduce latency for global users, and provisioning high throughput to handle traffic efficiently during peak usage. These considerations align with the cost tradeoffs and performance requirements of AWS generative AI services.
- A. Correct.
Token-based pricing is cost-effective because it allows the company to pay only for the amount of data processed, which helps control costs.
- B. Correct.
Deploying the model in multiple AWS Regions ensures low latency for users across the globe, improving responsiveness and availability.
- C. Incorrect.
Using a pre-trained foundation model might not meet the company's needs for a custom chatbot, as it limits their ability to tailor the model to their specific requirements.
- D. Correct.
Provisioning high throughput ensures the model can handle high traffic during peak demand, maintaining responsiveness for users.
- E. Incorrect.
Deploying the model across Availability Zones in a single region improves redundancy locally but does not address the need for global availability and responsiveness.