Meta’s plans for generative AI are starting to take shape and although the company is far from Microsoft’s goals with Bing Chat, Mark Zuckerberg has made it clear that he wants Meta to be considered one of the leading companies in the industry.
According to several sources, during a company meeting with staff this week, Zuckerberg outlined some of Meta’s plans in more detail. The CEO told employees some of the ways Meta intends to incorporate generative AI into each of its products. The planned artificial intelligence features will be introduced in WhatsApp and Messenger in the early stages, closely followed by the introduction of a chatbot within Instagram. This feature has emerged recently thanks to expert Alessandro Paluzzi, who spotted the first signs, indicating that the app may have up to 30 different “personalities” to choose from.
Further rumors say that a generative AI photo editing feature is in the works on Instagram. The feature would allow users to edit their photos through text prompts and then share the images in Stories. Zuckerberg also recently discussed post-creation tools for Facebook, as well as the platform’s advertisers, at the company’s earnings conference.
As for the introduction in Messenger, during the same company meeting mentioned above, Ahmad Al-Dahle, vice president of AI at Meta, told employees that the company will use its image generation model to allow users to create stickers based on text prompts. Employees will begin testing the feature internally before it is made available to the public. We also have a quotation mark shared by The Verge and coming from Ahmad himself.
“With the stickers generated by the I, our users can have endless options to express themselves, cultural representations and even the relevance of trends. Of course, stickers are just the tip of the iceberg. AI models that will transform any image you want the way you want”
Al-Dahle adds that the company is also working on AI models that will transform any image the way you want and that will include the ability to change the aspect ratio of an image, turn an image of a corgi into a painting, and many other possibilities not shared during the meeting.
The company is also delving into AI with its LLaMA language model and recently announced a new open-source AI model called ImageBind, which combines textual, audio, visual, motion, thermal, and depth data. To what end? Such a powerful technology could find use in virtual reality, or even in the real-time generation of complex environments or simulated life situations, but we are really talking about science fiction for now.
