AWS Bedrock—Amazon’s Generative AI Launch
By: Craig Kennedy
AWS Bedrock—Amazon’s Generative AI Launch
Today, AWS announced that it is entering into the generative AI market that has been dominated lately by competing announcements from Microsoft and Google. These new capabilities, accessible through the newly announced Amazon Bedrock service, are currently in limited preview.
Amazon Bedrock
Amazon Bedrock is a new AWS service that provides their users access to generative content foundation models (FM) through a set of APIs. Users can select from three popular 3rd party FMs like AI21labs Jurrasic-2, Anthropic Claude, or Stability.ai Stable Fusion, or they can choose a FM from a new family of AWS developed FMs collectively named Amazon Titan.
The exciting part of the Amazon announcement is that Amazon Bedrock enables fine-tuning of the models by training them with the customer’s own proprietary data while ensuring that the data is kept private and secure.
Amazon Titan
Amazon Titan is a new family of AWS developed foundation models (FM) that will be available to AWS users through the Amazon Bedrock service. Amazon currently has two large language models (LLM) available in Titan.
Titan Text is a generative LLM capable of classification, text generation, information extraction, summarization, and open-ended Q&A. Titan Embeddings is a LLM that converts text input into numerical representations that persists the semantic meaning of the text in a much more efficient way.
Titan FMs are designed to detect and remove harmful content in the data, reject inappropriate content in the user input, and filter model outputs that contain inappropriate content.
Amazon CodeWhisperer—General Availability
Amazon also announced the general availability of Amazon CodeWhisperer. Amazon is also making the individual tier of the service available to all developers for free. CodeWhisperer can be integrated to many of the popular developer IDEs, or natively in AWS Cloud9 or AWS Lambda console, and supports a wide range of programming languages.
Two of the standout features in CodeWhisperer are its Code Reference Log and Security Scan capabilities.
Code Reference Log provides the ability to identify and either flag or filter code that resembles open-source training data and displays the open-source project’s repository URL and license, letting you easily review and add attribution.
Security Scan will scan your code (both developer-written and generated) for vulnerabilities and provide suggestions to remediate them. It scans for vulnerabilities, such as those outlined by Open Worldwide Application Security Project (OWASP), or those that don’t meet crypto library best practices and other similar security best practices.
Bottom Line:
AWS had been notably silent over the last several months as the other large cloud providers have been making very visible announcements. It appears that they have been working diligently with their heads down on this project building up to this announcement.
We’ll see what is finally delivered when these announced offerings are generally available, however based on the features being touted by Amazon they appear well thought out and can be a powerful addition to the other generative content AI solutions currently available.
It’s all about AI in our next Transform Tour Stop!
Join us for #TransformTour23, AI Everywhere-Going ChatGPT, on Thursday, June 15th at 10 AM PT/ 1 PM ET. Register for Free Here!
This blog is a part of the Digital Operations blog series by Aragon Research’s Sr. Director of Research, Craig Kennedy.
Missed an installment? Catch up here!
Have a Comment on this?