Amazon Uses Multiple AI Models

Amazon is pursuing several artificial intelligence efforts, each apparently with a different monetization model. Some models are intended to support Amazon’s own e-tailing efforts, while others are aimed at third-party enterprise customers buying cloud computing capabilities. 

Among the latest efforts is a new large language model, codenamed Olympus, intended to be sold to AWS cloud computing customers and obviously pitched as a competitor to Google and Microsoft LLMs. 

Olympus is expected to perform better than Titan, another group of Greek-named LLMs that AWS is currently selling to cloud customers. Up to this point, AWS has been selling the use of Titan to develop applications with personalization and search capabilities. 

Two other Titan models, designed to help customers create apps that generate text or summarize long tracts of text do not appear broadly available to AWS customers, some note. 

AWS also has been supporting LLMs developed by Anthropic, as well as other models customers might prefer, such as the AI21 Labs LLM. 

But Amazon also has been using AI to support its own retail operations as well. 

There are several key e-commerce use cases.. Amazon monetizes its product recommendations by increasing sales.

Amazon also monetizes its AI-powered search results by displaying sponsored products at the top of the search results page. Advertisers pay Amazon to display their products in sponsored search results.

Amazon monetizes its AI-based fraud detection by preventing financial losses.

Amazon monetizes its inventory management by using AI to reduce costs while Amazon monetizes its logistics network by reducing costs and increasing sales by charging partners to use the network. 

AWS has different monetization avenues, all based on offering cloud computing services to enterprises and other customers. For example, AWS monetizes its machine learning services by charging customers for usage.

Natural language processing services can be purchased by AWS customers as well. Customers are charged based on the number of API calls that they make and the amount of data that they process. So that particular AI service earns revenue for AWS based on usage. 

Likewise, AWS monetizes its computer vision services by charging customers for usage. Similarly, AWS monetizes its speech recognition and synthesis services by charging customers for usage.

Amazon monetizes Alexa by selling Alexa-enabled devices and services. Amazon also generates revenue from advertising on Alexa devices and services. Amazon also monetizes Amazon Robotics by selling robots to other companies.

All of those features and services are forms of AI-enabled products. 

Name	Target Audience
Megatron-Turing NLG	Researchers and developers working on natural language processing (NLP) and machine learning (ML) tasks, such as text summarization, question answering, and machine translation.
Amazon Comprehend	Businesses and developers who need to extract meaning from text data, such as customer reviews, social media posts, and product descriptions.
Amazon Rekognition	Businesses and developers who need to extract meaning from images and videos, such as identifying objects, detecting faces, and understanding scenes.
Amazon Transcribe	Businesses and developers who need to transcribe audio to text, such as customer calls, meetings, and lectures.
Alexa	Consumers who use Alexa-enabled devices to control their homes, get information, and access entertainment.

--------------------