February 23, 2024

An Amazon Fire TV Omni QLED displaying an image created by the prompt, “Alexa, create a picture of a hidden temple in a rainbow jungle.”


Companies big and small are adding generative AI features to different technologies and devices to give them an extra boost. But even though these AI features are spreading like pollen in April, one company stands out for how they’re using it: a voice-to-image generator built into a smart TV.

Amazon is finally rolling out its AI Art feature for Fire TVs in public preview, which lets users ask Alexa for an image and then watch as their imagination comes to life and fills the TV screen. Users can then add this image to their Amazon Photos account, display it as the background for their Fire TV, or discard it. 

Also: Alexa just got three new generative AI skills – here’s how to try them

The caveat? The AI Art feature is only available in the latest-generation Fire TV Stick 4K Max and the Fire TV Omni QLED series as part of Amazon’s Ambient Experience, which is only available on these devices.

Users that have either of these devices can activate the AI Art feature by enabling the Ambient Experience on their TV and agreeing to the AI Art public preview. Then, it’s just a matter of asking Alexa for an image and choosing from the four generated outputs.

Like other AI image generators, users can give prompts that are as detailed as they’d like, even including the artwork style they’re looking for, like a cartoon, painting, photo, 3D render, watercolor, and more. Users can also add art movements to their prompts, like impressionism, Cubism, or Pop Art. 

Also: ‘Generative AI by iStock’ lets users create images without copyright-infringement worries

The idea behind the AI Art feature is for consumers to create their artwork for their Fire TVs using just their voice. This generative AI feature is powered by both Alexa for voice prompts and Amazon’s Titan Image generator to create the images using AI. 

Amazon’s Titan Image Generator is a proprietary vision-language model that generates images from text prompts, such as spoken text. The generative AI model is still in preview and has been trained on carefully curated photos and captions to produce realistic images. 

Also: Windows 11 Paint has its own AI image generator tool – here’s how to use it

Regarding copyright concerns for AI-generated art, an Amazon spokesperson told me, “Titan models are trained on data from a variety of sources, including licensed data, open source datasets, and publicly available data where appropriate.”

While this doesn’t ensure the generation of commercially safe images, the Titan Image Generator does add a digital watermark to its images to flag them as synthetic content. This doesn’t speak to copyright concerns directly but suggests an awareness of the need to distinguish AI-generated images from human creations.