AWS - three-layer approach to AI

AWS has developed a comprehensive three-layer approach to artificial intelligence and machine learning, particularly in the context of generative AI. This approach is designed to cater to different skill levels and use cases across organizations.

The bottom layer focuses on infrastructure and tools for building and training large language models (LLMs) and foundation models (FMs). This layer includes high-performance computing resources like NVIDIA GPUs, custom machine learning chips such as AWS Trainium and Inferentia, and services like Amazon SageMaker that simplify the machine learning development process.

The middle layer is centered around providing access to models and tools for building and scaling generative AI applications. A key component of this layer is Amazon Bedrock, which allows developers and data scientists to create secure, customized, and responsible generative AI applications using LLMs and FMs from leading AI companies.

The top layer consists of AI-powered applications that leverage LLMs and other foundation models to enhance productivity and streamline various tasks. This includes tools like Amazon Q, a generative AI-powered assistant that can accelerate software development, generate content, derive insights, and take actions based on a company's internal data.

This three-layer approach allows AWS to provide AI and ML capabilities suited for different needs, from ML experts and practitioners working with raw infrastructure, to data scientists and developers using pre-built services, to end-users who can benefit from AI-powered applications without needing to build models themselves.

AWS emphasizes security across all three layers, leveraging technologies like the AWS Nitro System to ensure the protection of customer data and workloads. This comprehensive approach enables organizations to adopt and implement AI solutions at various levels of complexity and scale, depending on their specific requirements and technical capabilities.
Sources
Powering the generative AI era: What you missed at the AWS Public Sector Symposium Brussels | AWS Public Sector Blog
Accelerating the next wave of generative AI startups | AWS Startups Blog
Artificial intelligence and machine learning (AI/ML) - AWS Prescriptive Guidance
A secure approach to generative AI with AWS | AWS Machine Learning Blog