Amazon Web Services (AWS) continues with its ascending line of growth, and in this case, with its clear commitment to the generative AI that drives the creation of AI platforms for developers and startups. This was stated at the summit held on July 25 in New York.
Remember that an internal document was transmitted in May announcing the creation of an AI team in AWS, with deepak singh As responsible. The new department called ‘Next generation experience for developers’ arises with the idea of transforming the way customers use AWS by incorporating generative AI.
In order to improve developer productivity, AWS recently introduced products such as bedrocka service that facilitates access to LLMs through an API, and codewhispereran AI programming assistant.
To build and promote a new AI model, it is necessary to real chips, but also a phase of training and its application to the real world. Only then will it be possible to compete against OpenAI’s ChatGPT.
AWS HealthScribe
The main announcement has been the launch of this generative service powered by AI suitable for hippa which is responsible for transcribe conversations between patients and doctors, creating efficient clinical documents. Includes AI-generated doctor-patient relationship summaries as well as structured medical terms.
AWS HealthScribe arises to reduce the time doctors spend writing documentation, gaining greater profitability and efficiency. A single API can now be used to automatically create transcripts, extract details, and produce clinical summaries for later insertion into an electronic health record (EHR).
However, the integration of AI in clinical applications poses significant challenges such as the implementation complexity (it is a laborious and expensive operation), the process safety (they must ensure that the strictest privacy requirements are met. The data is deleted and encrypted after processing) and the trust (vital for training new models).
AWS HealthScribe combines conversational and generative AI to reduce the burden of clinical documentation and improve the patient experience. It is designed as assist functionin such a way that each summary generated by AI is linked to the transcript of the consultation, allowing users to easily verify their clinical data with full transparency and traceability.
It is not the first time that AI is oriented towards the medical sector, since last April Google launched Med-PaLM 2.
Accessible Generative AI Courses
It is another of the great novelties presented by AWS. have been incorporated seven generative AI courses for people with different skills and work experiences, including aspects as complex as learning to build with Amazon CodeWhisperer or learning different methods of applying AI in business.
With Amazon Bedrock developers can choose the model they want for each case, including Titan from Amazon, claude from Anthropic, stable diffusion from Stability.ai and Jurassic-2 of AI21 Labs, until today. You can also access claude 2Anthropic’s latest LLM, SDZL 1the latest Stability AI text-to-image model and a new model, Cohere.
Generative AI becomes more accessible, and developers can build apps that can perform a myriad of tasks with up-to-date responses. However, the new features of Amazon Bedrock are still in progress. preview version.
Vector Engine for Amazon OpenSearch Serverless
The new AWS vector engine for Amazon OpenSearch Serverless it makes it easy for developers to find embeds and later incorporate them into LLM applications.
Available in preview, it allows developers to store, search and retrieve billions of vector embeds in real time, neglecting the underlying infrastructure. This will allow you to create modules that produce better and more accurate results.
Amazon EC2 P5 Public Availability
What was announced by Amazon in March has been confirmed. Amazon Elastic Compute Cloud (Amazon EC2) P5 instances powered by Nvidia H100 Tensor Core GPU are available to the general public. This new tool optimizes the computational performance required to create and train ML models. With it, training times are up to six times faster than previous models and it is allowed reduce training costs by up to 40%.
AWS Economic Balance
The exponential growth of Amazon’s cloud computing platform is seen in its figures, since during the first quarter of 2023 they registered an improvement of 16% compared to the growth of 9% obtained by Amazon in general.
AWS assumes $21.4 billion in sales for Amazon. However, the net profit suffered variations and went from 3,800 million in 2022 to 3,200 million dollars.
