Defiance ETF CEO and CIO Sylvia Jablonski is responding to reports that Amazon and Google are tempering expectations for generative AI to “make money.”
Amazon CEO Andy Jassy outlined the e-commerce giant’s approach to artificial intelligence (AI) development in his annual letter to shareholders released Thursday.
Jassy wrote that Amazon aims to “empower internal and external builders” in its approach to developing products and services for its customers. He goes on to characterize builders as “people who like to invent” and go through a process of iterative improvement to enhance existing tools and experiences.
“It’s important for builders to have the right tools to continue improving the customer experience quickly,” he writes. “The best way to accomplish this is to build primitive services. Think of services as discrete building blocks that can be assembled in any combination the builder wants.”
In the context of generative AI (GenAI), Jassy explained that Amazon and AWS are applying this approach to their AI efforts to meet a variety of needs.
AWS rolls out AMAZON Q, an AI assistant for business
Amazon CEO Andy Jassy emphasized the company’s AI push in a letter to shareholders. (Kevin Winter/Getty Images/Getty Images)
“Much of the early public attention was focused on GenAI applicationWith the notable release of ChatGPT in 2022,” he said. “But in our ‘primitive’ way of thinking, the GenAI stack has three different layers, each of which is huge, and each of which we are deeply invested in.
“The bottom layer is for developers and companies who want to build a foundational model (“FM”). The key primitives are the computers required to train models and generate inferences (or predictions) and the software that facilitates the construction of these models. ” he explained.
“To date, virtually every major FM has been trained on Nvidia chips, and we continue to offer the broadest collection of Nvidia instances of any provider,” Jassy said. “However, supply is in short supply and cost remains an issue as customers expand their models and applications.”
Jassy said that in response to customer requests that Amazon “push the boundaries of price/performance for AI chips,” similar to what it did with its Gravitron general-purpose CPU chip, the company is launching a custom AI training chip called Trainium and an inferencing chip called Inferentia. He writes that he has developed a chip.
AWS launches program to increase AI and technology capabilities for small and medium-sized businesses
Amazon Q AI tools help businesses analyze their data. (Jaap Arrians/NurPhoto via Getty Images / Getty Images)
Last fall, he said, Anthropic, a leading FM manufacturer, announced that it would use chips from Trainium and Inferentia to build, train, and deploy these models, and those chips are now being rolled out to a second generation with improved cost and performance. He said that it has become a version.
“The middle tier is for customers who want to leverage their existing FM, customize it with their own data, and build GenAI applications with the security and capabilities of a leading cloud provider, all as a managed service.” Jassy said.
“Amazon Bedrock invented this layer to make it the easiest way to build and scale GenAI applications with a wide selection of first- and third-party FMs and easy-to-use features that give GenAI builders access to high-quality models. Provide customers with faster output.
“The top layer of this stack is the application layer. We are building a significant number of GenAI applications across all of Amazon’s consumer businesses.”
What is artificial intelligence (AI)?
Last year, Amazon released Rufus, an AI-powered shopping assistant. (Photo illustration: Sheldon Cooper/SOPA Images/LightRocket, Getty Images / Getty Images)
Jassy explained that these GenAI apps include a new AI-powered shopping assistant, Rufus, which is a more intelligent version of Alexa. Advertising feature that allows users to generate images. Video and copy using natural language. Productivity app for customers and sellers.
The company is also building applications within AWS, including what Jassy called “coding companions, which is probably the most compelling early GenAI use case.”
“We recently wrote, debugged, tested, and implemented code while also performing transformations (e.g., migrating from an older version of Java to a newer version) and using our customers’ various data repositories (e.g., intranets, wikis, etc.). , Salesforce, Amazon S3, ServiceNow, Slack, Atlassian, etc.) to answer questions, summarize data, keep a consistent conversation going, and take action,” he explained. “Q is the most capable work assistant available today, and it is rapidly evolving.”
CLICK HERE TO GET FOX BUSINESS ON THE GO
Jassy said AWS’ AI efforts to meet these diverse customer needs will help drive the next phase of AI development both internally and externally.
“These AWS services are comprised of a set of building blocks that democratize the next creative phase of AI at all three layers of the stack, enabling internal and external builders to build on what we know and understand. We are optimistic that much of this world-changing AI will be built on AWS. ,” Jassy wrote.
