Amazon Payments has implemented a multi-objective contextual multi-armed bandit model on Amazon SageMaker to personalize content for customers. This approach aims to optimize the selection of personalized content generated by generative AI, addressing the challenge of choosing the best option for each user. In a seven-week test, the system showed a high single-digit percentage lift in conversion for one customer group, though another group saw no improvement, indicating the content itself was the limiting factor. AI
IMPACT Enhances AI-driven personalization by optimizing content selection, potentially improving customer conversion rates.
RANK_REASON Article describes the application of an AI method (contextual bandits) to a specific product/service (personalization on AWS) by a company (Amazon Payments).
Read on AWS Machine Learning Blog →
- Amazon Bedrock
- Amazon SageMaker
- AWS
- contextual bandits
- generative AI
- Thompson sampling
- University of California, Berkeley
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →