Amazon announces new AI models

By Greg Bensinger

LAS VEGAS (Reuters) – Amazon announced a new set of artificial intelligence platforms, called Foundation Models, that enable the generation of text, images and videos, among other things, at its annual AWS conference.

The new offerings pit the company against rivals like Adobe and Meta, which are vying to serve customers who want to automate their services more.

Amazon boss Andy Jassy announced the new “Nova” models on Tuesday. Developers, Jassy said, had a list of wishes that provided the impetus for the new services.

“They want better latency. They want lower costs. They want the ability to fine-tune,” Jassy said.

The announcements at the Las Vegas conference are Amazon’s biggest step in combating a reputation that the company was caught on the wrong foot in developing AI applications as competitors moved faster.

Video generation from a single image or text prompt is particularly popular as Adobe, Meta, OpenAI and Tik Tok parent company ByteDance, among others, have announced new AI applications of this type. Amazon announced Tuesday that its Nova Reel software allows users to create six-second videos that can be useful for things like displaying products on Amazon’s website. Amazon says videos up to two minutes long will be available in the coming months.

Entertainment industry technologists are eager to get their hands on such tools to improve and accelerate film production capabilities more efficiently. Still others fear that such systems could infringe copyrighted works.

Also on Tuesday, Amazon announced that it had developed Canvas to generate images from short text prompts. Jassy emphasized that Amazon would introduce watermarks to ensure that the software is used responsibly and prevent the spread of harmful content.

Further offers announced on Tuesday are intended to speed up the processing and analysis of texts.

(Reporting by Greg Bensinger in Las Vegas; Editing by Will Dunham)