Amazon's New AI Offerings

By Greg Bensinger
LAS VEGAS (Reuters) – Amazon introduced a series of artificial intelligence platforms, foundation models, at its annual AWS conference, enabling text, image, and video generation, among other functionalities.

The new models position Amazon against competitors like Adobe and Meta, both of which are striving to meet customer demands for automation in their services.

Amazon's CEO, Andy Jassy, unveiled the new "Nova" models on Tuesday, highlighting that developer requests influenced these new services.
Jassy noted, "They want better latency. They want lower cost. They want the ability to do fine tuning."

These announcements signify Amazon’s significant move to enhance its reputation in the AI space, particularly after concerns that it lagged behind competitors in development.

Video Generation: A trending area, video generation from single images or text prompts is hot, with firms like Adobe, Meta, OpenAI, and ByteDance launching similar applications. Amazon's Nova Reel software enables users to create six-second videos, useful for showcasing products on the Amazon website, with capabilities for longer two-minute videos coming soon.

Entertainment industry innovators are keen to utilize these tools for more efficient filmmaking processes. However, there are concerns about potential infringement on copyrighted materials.

Image Generation: On the same day, Amazon introduced Canvas, a tool for generating images from brief text prompts. Jassy highlighted that watermarking would be implemented to promote responsible use of the software and prevent the spread of harmful content.

Additional tools were also introduced to enhance the speed of text processing and analysis.