At Amazon re:Invent in Las Vegas today, Amazon Web Services (AWS) CEO Andy Jassy introduced three new artificial intelligence and machine learning services to help developers create cool new applications based on images, text, and audio.
Jassy said a lot of companies don’t realize the heritage Amazon has in artificial intelligence (AI) and machine learning. He said Amazon has had “thousands of people dedicated to AI” for a long time. Their work has resulted in Amazon’s recommendation engines for products. The company’s fulfillment processes are also driven by machine learning. And its Alexa service that powers its Echo devices is based on Amazon AI software.
Amazon not only wants to use its AI capabilities for itself, but it also wants to sell its knowledge-base to others who want to develop new applications.
“We have always tried to take and make what we use ourselves and expose it for you,” said Jassy.
To that end, he announced Amazon Rekognition, a new service where software detects and categorizes objects in an image. It allows customers to build applications that can search for data within pictures. And it goes beyond the basic facts in the image to also include context such as whether a person in the picture is smiling or frowning.
Secondly, Jassy announced Polly, a text-to-speech service. Users can send a text message to Polly, and the software then spins out an MP3 stream that repeats and clarifies the message. It’s available in 47 different voices in 24 languages.
AWS also capitalized on the natural language platform it uses for Alexa, today rolling out Amazon LEX. The natural language service allows developers to create conversational applications. For example, an end user might submit a piece of text or audio to inquire about booking a flight. Then the app could ask relevant questions – kind of like a travel agent – to find and book the best flight.
New EC2 Instances
Jassy also today introduced a slew of upgrades to its EC2 instance families.
AWS T2 instances are its low cost, versatile instances. Today the company launched two more sizes within T2 to meet customer demand for larger memory footprints.
AWS introduced R4 instances, expanding on its R3 memory-intensive family. In R4, it boosted both memory caching and memory speed.
Expanding on its I2 instances, AWS announced that in 2017 it will roll out I3 instances designed to meet the needs of the most demanding I/O workloads.
Boosting its C4 compute-optimized instances, AWS announced its new C5 family based on Intel’s new Xeon Skylake processor.
To give AWS customers access to hardware acceleration, the company is making programmable hardware — field-programmable gate arrays (FPGAs) — available to its F1 instances.
For customers who want to spin up virtual private services (VPS), quickly, AWS created its new Lightsail VPS service.
Finally, Jassy said customers don’t always need all their available GPU cycles; sometimes they’re running a workload where they need “just a sliver of a GPU.” So today, he announced a new service called elastic GPU. It allows customers to attach GPU to any of AWS’s nine EC2 instances.
Photo: AWS CEO Andy Jassy at Re:invent, talking about how businesses must embrace technological change if they want to achieve “immortality.”