Meta has yet to give clear signals regarding its willingness to shift the focus from virtual reality to artificial intelligence. During the last call on the company’s earnings, Mark Zuckerberg spoke in detail about this need without going into the smallest details. Thanks to past interviews and statements made to investors and the press, not only by the CEO of Meta but also by his colleagues, we can already frame the 5 most interesting uses of generative AI between Facebook and other proprietary services.
Generative Advertising
Facebook, Meta’s largest platform and the world’s largest social network, will turn its gaze towards AI to create increasingly intriguing advertisements capable of attracting new consumers. Therefore, generative artificial intelligence could become one of the most used tools by advertisers and large companies to develop ads in multiple languages, colors and with different texts according to the target audience, dividing the target by age, interests or region in which they reside. In short, the generation of advertisements could become immediate and intuitive, leaving more freedom to advertisers and making their work easier.
AI-Powered Generative Chat
A Meta-signed chatbot must exist, given the boom of ChatGPT and Bing Chat. However, Zuckerberg spoke of the “next level”, namely “chat experiences in WhatsApp and Messenger with artificial intelligence agents”. In simpler terms, the US giant could create more AI characters ready to help people in various ways, whether writing code, writing creative texts, technical assistance or automated customer service. However, implementing these services will take a long time, given the need to renovate the data centers and build an infrastructure that can allow the correct operations of the AIs following the international rollout.
Image Generation
Meta’s Facebook AI division has developed its image generation technology called Instance-Conditioned Generative Adversarial Networks ( IC-GAN ). The researchers say that its use could prove very effective for creating different images than those traditionally seen with Midjourney or Stable Diffusion, as it could exploit a smaller amount of data from the real world to synthesize a richer set of data, reducing the cost of generating, collecting and archiving documents for algorithm training. Also in the company’s plans is a text-to-video generative AI application, Make-A-Video, to be embedded on Instagram Reels.
Natural Language Generation
The LLaMA – Large Language Model Meta AI – language model is already known, i.e. the direct response to ChatGPT and Google Bard. Deliberately designed as a smaller model, trained on “just” 65 billion parameters, LLaMA would require less computing power and allow Generative AI to run on smaller systems, wasting fewer resources than the cloud servers needed for ChatGPT or Bard. Indeed, there is even talk of the possible autonomous execution on PCs and smartphones. The quality of the outputs may be lower, but there will be no hits in the functioning, or at least, hopefully.
The AI In The Metaverse By Meta
Naturally, artificial intelligence must be included in Horizon Worlds and, generally, in the metaverse. After all, it’s the reason Facebook changed its name. The “next generation” of the Internet between virtual reality (VR), avatars and the sharing of online experiences could be equipped with advanced AI tools, fueling the creativity of the Horizon Worlds proprietary platform itself and allowing the creation of worlds, games or other activities in a short time, for fans around the world.
Also Read: How The Metaverse Will Change The Workplaces Of The Future