By Vikas SN
Facebook parent Meta is infusing generative artificial intelligence (AI) into its applications – Instagram, WhatsApp, and Messenger – used by over 3 billion people daily, as the social networking giant jostles to be a prominent player in the AI arms race with tech giants, such as Google, Microsoft, and upstarts such as OpenAI.
Generative AI is a form of artificial intelligence that can generate text, images, sounds, and other media formats in response to short commands or prompts from users.
“This is really just the beginning. If you look across the industry and at what everyone is doing, most people haven’t yet had the chance to experience any of these AI advances yet. And, that’s the thing that we can help change” Meta chief executive Mark Zuckerberg said during his keynote address at the company’s annual Connect developer conference on September 27.
Meta AI
Meta is introducing a new AI chatbot called Meta AI that will be soon available on WhatsApp, Messenger, Instagram, and later on Quest 3 mixed reality headset and Ray-Ban Meta smart glasses.
Powered by a custom AI model that leverages technology from Meta’s open-source large language model (LLM) Llama 2 and the company’s latest LLM research, Meta AI acts like a general-purpose assistant that can answer a variety of questions, including real-time information through a search partnership with Microsoft’s Bing, as well as generate realistic images.
This assistant can be invoked by users in personal and group chats across its applications. For instance, one can ask for recommendations for a group trip, or ideas for dinner party recipes, following which it will surface various options directly in the chat. One can also type “@MetaAI /imagine” followed by a descriptive text prompt to create photorealistic images within seconds.
“We are going to focus over the coming months and years on building this into a deeper personal assistant with more integrations across everything we do,” Zuckerberg said.
Over the next month, applications such as WhatsApp, Messenger, Instagram, and Facebook Stories will let users type in a description to generate custom AI stickers on the fly and send them to their friends. This feature uses technology from Llama 2 and the company’s foundational model for image generation called Emu. It will initially be available to select English language users.
Meta is also introducing AI editing tools – Restyle and backdrop – to Instagram next month. Restyle lets users reimagine their images by applying a new visual style that they describe while backdrop lets them change the scene or background of the image by creating an image of the primary subject in the foreground with the background described by the user.
The company noted that images created with these tools will indicate that it was an AI-generated image and they are experimenting with various forms of visible and invisible markers.
AI Characters
The social networking giant is also introducing an initial set of 28 character-based AI chatbots across its messaging apps that would have more personality, opinions, and perspectives on topics such as travel, games, and food, but would only answer questions.
These AI characters are currently limited to information that largely existed prior to 2023, which could mean some responses may be dated. However, the company said it is planning to bring search capabilities to many of these chatbots to provide timely and relevant conversations.
Meta has partnered with various cultural icons and influencers to play and “embody” some of these AI characters, which will each have a profile on Instagram and Facebook. Among them include social media personality Charli D’Amelio, supermodel Kendall Jenner, YouTube star MrBeast, tennis player Naomi Osaka, socialite Paris Hilton and rapper Snoop Dogg.
“This isn’t just going to be about answering queries. This is about entertainment and about helping you do things to connect with the people around you” Zuckerberg said. For instance, while chatting with one of the AI characters, their profile images will change expressions based on the conversation.
Meta said that these celebrities will be representing the characters, not themselves and the public-facing social profiles and related visual content creation will be managed by them. It will be limited to text chats at present with plans to add voices early next year.
These AI characters are available in the United States starting today and the company plans to add more characters including British adventurer Bear Grylls in the coming weeks.
“One of the views that we have on the development of AI, that differs from the rest of the industry, is that we don’t think that there will be a singular super intelligence that everyone interacts with. Our view is that people will want to interact with a bunch of different AI (chatbots) for different things that you want to do” Zuckerberg said.
AI Studio
The Meta chief also unveiled AI Studio, a platform for people to create these types of AI characters. Developers will be able to build third-party AI characters for the company’s messaging services with their APIs in the coming weeks, starting with Messenger and later expanding to WhatsApp.
Businesses will be able to create AI characters that reflect their brand’s values and improve customer service experiences across their apps. Zuckerberg said they are launching it in alpha with select businesses today and will roll out more broadly next year.
In the first half of next year, creators can build these AI characters to extend their virtual presence. Meta said these characters will have to be sanctioned by them and directly controlled by the creator.
“We’re also building a sandbox so that people who don’t code can also train AI (characters). We hope to have that out sometime next year. And we’re also working on bringing all of this to the metaverse where you are going to be able to have these embodied as avatars” Zuckerberg said.
He said that they are rolling out these AI tools and features slightly slower than they traditionally do since they want to ensure it is done thoughtfully and they can iron out issues before these features get to scale. “Generative AI is powerful and there is a lot of awesome potential, but there are also going to be some new challenges that come with this,” he said.
The company has been training and fine-tuning the AI models to fit their safety and responsibility guidelines as well as brainstorming on the potential challenges and building as many safeguards as possible, Zuckerberg said.
“We’ve done a lot of red teaming with experts on this and we’ve been building guardrails around inappropriate conversations” he added.
Quest 3 headset and Ray-Ban Meta smart glasses
At the event, Zuckerberg also unveiled Quest 3, the company’s first mixed-reality headset that blends in the physical and virtual worlds together. Earlier versions of Quest headsets were focused only on virtual reality.
The Quest 3 is currently available on pre-orders and will start shipping on October 10. Pricing starts at $499.99 for the 128GB version and goes up to $649.99 for the 512GB variant.
This launch comes after tech giant Apple entered the fray with the launch of its mixed reality headset, Vision Pro, in June 2023. The headset however has a hefty price tag of $3,499.
In addition, the Meta chief executive also introduced a new generation of Ray-Ban Meta smart glasses, in partnership with EssilorLuxottica. These glasses will come built-in with Meta AI assistant and will allow users to livestream from the glasses to Facebook or Instagram.
It is currently available for pre-order on Meta and Ray-Ban websites with prices starting at $299. It will be available in 15 countries, including the United States, Canada, Australia, and Europe, for online purchase and in retail stores starting October 17.