The past year has seen remarkable advances in the field of artificial intelligence (AI) by Meta. The company has introduced new AI experiences in its applications and devices, highlighting significant achievements such as the Llama family of language models and research breakthroughs such as Emu Video and Emu Edit. These achievements promise to unlock revolutionary capabilities in its products for the coming year. Last Wednesday, the company announced key updates to its core AI experiences and new capabilities that will enrich the user experience across its various applications.
Imagine with Meta
With just days to go before the end of 2023, Meta pulls out an ace up its sleeve and surprises us again by launching a new standalone AI tool called “Imagine with Meta”, with which it is possible to generate images from text in everyday language.
The tool, which uses Meta’s Emu image generation model, is capable of generating high-resolution images from everyday language prompts. At the moment, Imagine is available to U.S. users, who can generate up to 4 images per prompt. Meta has not yet indicated when it will roll out the tool to the rest of the world, so for now we will have to settle for using a VPN to test it in Spain. If you wish to try out this new tool, all you have to do is go to the Imagine with Meta website.
This is not Meta’s first approach to AI. In fact, it had an AI-powered sticker generator, but in addition to the fact that it will not allow users to generate images, it was accused of being racially biased, which many worry about the precautions taken with Imagine to avoid these same problems.
While Imagine is not adding watermarks at the moment, the technology company has stated that all images generated in the coming weeks will have a watermark in the lower right corner, indicating that the image has been created with Meta’s AI.
Although the tool was not subjected to any live testing before being launched, Meta is committed to ensuring that the images can be correctly identified. In fact, the new watermark will be invisible and indelible, remaining even if it is cropped, resized, colored or screenshotted.
In this way, Imagen joins the growing ecosystem of image-generating AIs, with a model similar to other models such as Bing Chat and Dall-E 3, Midjourney, Stable Diffusion, Firefly and other technologies.
A flurry of Meta AI-related news
Meta AI, the company’s virtual assistant, has evolved to provide more detailed answers and accurate summaries of searches, improving the user experience on mobile devices. This development translates into more useful answers for a wider variety of requests. Interaction with Meta AI has been simplified, allowing users to access through messaging and messaging platforms, as well as through voice commands with Ray-Ban Meta smart glasses.
The large language model technology powering Meta AI will be used to suggest comments on posts, group chat topics, provide search results and even enhance product descriptions in stores, introducing the new standalone experience called “imagine with Meta AI” for creative enthusiasts.
New experiences with Reels in Meta AI
The inclusion of Reels in Meta AI chats offers users a new way to discover visual content, connect with creators and find inspiration. This integration is an early example of the deeper connections the company plans to implement in its applications.
Improving the Facebook experience
Meta AI also plays a role in improving the Facebook experience, from creating personalized birthday greetings to suggesting groups and editing posts in the Feed.
Helping creators respond to their fans
The introduction of suggested replies in direct messages seeks to streamline interaction, allowing creators to get more engagement with their followers.
Chatting with AIs
Meta is evolving not only with Meta AI, but also with other AI-powered chatbots, such as Bru, Perry, Luiz, Coco, Lorena, Tamika, Izzy and Jade. The expansion of search and the introduction of long-term memory for some AIs offer users the ability to revisit past conversations, providing a more personalized experience. Each of these AIs will be specialized in a field, giving users the ability to have more productive conversations based on whether they are talking about sports, music, movies, etc. These enhancements are already available in the US.
Meta has made remarkable advances in artificial intelligence in the last year, and these developments not only improve experiences within its applications, but also set a standard for transparency and safety in AI content generation. As the company continues to explore the possibilities of AI, it is clear that we are still in the early stages of discovering its full potential. We will be watching for more exciting updates in the coming year.