AWS Showcases Generative AI-Ready Cloud Infrastructure

AWS is doubling down on its bid for the growing market for generative AI tools and infrastructure. During AWS re:Invent 2023 on Wednesday, the cloud hyperscaler highlighted its AI-ready platform that supports multiple large language models - and a new offering to compete in the crowded AI image generator space.

See Also: OnDemand | AI-Driven Endpoint Security: Adapting to Industry Changes

While the news about new AI offerings and partnerships has dominated the event, AWS executive Swamy Subramanian reminded attendees that the stakes are high and the future is riding on a symbiosis of human ingenuity and AI disruption in the years ahead.

"We're heading into a profound transformation where human-led creativity harnesses the transformative capabilities of gen AI, thus altering the trajectory of humanity," said Subramanian, vice president of data and machine learning.

AWS released the Amazon Bedrock managed service in September to support generative AI deployments for large enterprises, and it now supports easy access to multiple large language models and foundation models from AI21 Labs, Anthropic, Cohere, Meta and Stability AI. Subramanian said AWS Bedrock offers "more choices of leading models, customization features, agent capabilities and enterprise-grade security and privacy in a fully managed experience."

Anthropic, which builds reliable, interpretable and steerable AI systems, has introduced Claude 2.1 to Amazon Bedrock. Claude 2.1 will offer a 200,000-token context window and better accuracy over long documents. It helps businesses process text-heavy documents such as financial statements and internal datasets. Claude 2.1 can summarize, perform Q&A or contrast documents, and much more.

Meta has also forged a partnership with AWS, making Llama 2 70B available in Amazon Bedrock. Llama 2 is a collection of pretrained and fine-tuned large language models ranging in scale from 7 billion to 70 billion parameters. Llama-2-Chat, based on fine-tuned LLMs, is optimized for dialogue and human interaction.

AWS also plans to offer a complete infrastructure offering that enterprises can use to build gen AI applications. In addition to other components, it will include a broad choice of vector databases including Redis Enterprise Cloud, Pinecone, Amazon Aurora and MongoDB.

Amazon Finally Launches Its Own Image Generator

AWS has finally released the Titan Image Generator to compete against a large number of similar AI tools. Available for preview to AWS Bedrock customers, the image generator can create new images from text descriptions or customize existing images. AWS is targeting industries such as advertising, e-commerce, and media and entertainment, looking to explore tools for creating studio-quality, realistic images. The model can understand complex prompts and generate contextually aware images with accurate object composition and limited distortions.

The image generator adds an invisible watermark on all images it generates - part of a pledge for responsible AI made to the White House this year. The watermark is designed to help reduce the spread of misinformation by providing a discreet mechanism to identify AI-generated images and to promote the safe, secure and transparent development of AI technology.

“We are among the first model providers to widely release built-in invisible watermarks that are integrated into the image outputs and are designed to be resistant to alterations,” Subramanian said.