Meta offers preview of generative AI products for its Facebook and Instagram platforms

That’s why we are announcing the launch of a Community Forum designed to generate feedback on the governing principles people want to see reflected in new AI technologies. Our Community Forum on Generative AI will be held in consultation with Stanford Deliberative Democracy Lab and the Behavioural Insights Team (BIT). Both organizations were partners on our launch of our Community Forum pilots last year. Snap, Google and Meta all see applications for generative AI as the technology takes off. Company executives also demonstrated a coming Instagram feature that can modify user photos via text prompts and another that can create emoji stickers for messaging services, according to a summary of the session provided by a Meta representative. This is the web version of Data Sheet, a daily newsletter on the business of tech.

Meta’s new model will be “several times more powerful” than the Llama 2 generative AI model that was released earlier this year, according to sources cited by the Wall Street Journal today. A significant advantage of Meta’s involvement in this space is that these tools will be free for consumers, thanks to Meta’s ad-supported business model. Most competing apps on the App Store offer limited AI editing features for free, eventually pushing users into subscriptions for full access to all features. This new model generates text-to-images at a state-of-the-art rate while utilizing five times less computing power than earlier transformer-based techniques. It maintains low training costs and high inference efficiency while combining the adaptability and efficiency of autoregressive models. As a causal masked mixed-modal (CM3) model, CM3leon enhances the capabilities of prior models by being able to produce text and image sequences dependent on arbitrary sequences of other text and image content.

China successfully launches a pilot reusable spacecraft, state media report

Meta is giving people the option to access, alter or delete any personal data that was included in the various third-party data sources the company uses to train its large language and related AI models. Generative AI could be another creative tool that has applications in ad products. For instance, it’s easy to envision marketers using text prompts to tell Meta’s ad manager to design an ad. And messaging is one of Meta’s fastest-growing ad products on WhatsApp, which topped 2 billion monthly users last quarter. Businesses are using messaging ads to send products and offers to consumers, and Meta said “click to message” ads are now at a $10 billion annual revenue run rate.

The executive also indicated that the capabilities could eventually extend into more complex formats like video. Other aspects of the development roadmap seemed contingent on how advertisers respond to the tech. Executives acknowledged the popularity of rival generative AI products Yakov Livshits that have already been on the market for months but suggested that the sophistication and scale of Meta’s ad infrastructure could set it apart. AI Sandbox is being trained on Meta’s internal data, along with information from public sources and data that the company licenses.

AI, Advertising, and Hyper-creation

The company said that these features are available to select advertisers at the moment and it will expand access to more advertisers in July. While the guitar strings on the songs felt real, they still felt, well, artificial. Having a solid open source foundation will foster innovation and complement the way we produce and listen to audio and music in the future. With even more controls, we think MusicGen can turn into a new type of instrument — just like synthesizers when they first appeared.

• Examples of generative AI applications include ChatGPT – the fastest-growing application of all time, as well as image creation tools such as Dall-E and Stable Diffusion.
• Meta has put AI front and center in recent months, viewing it as a central pillar to growing its TikTok clone Reels and its longer-term metaverse ambitions.
• A spokesperson for Meta confirmed to TechCrunch that the company’s initial consumer-facing tools will be available this year, likely within the next few months.
• DALL-E translates text prompts into images, and ChatGPT writes fast responses to complex queries, like a more dynamic search engine.
• The brand also uses many of Meta’s “Advantage” AI- and machine learning-powered automated advertising tools, and Plofker is a big proponent of the potential for generative AI to further streamline marketing processes.

Participants will explore the principles a diverse range of users from around the world believe generative AI systems should align with. Our hope is that the forum will not only help inform our own modeling infrastructure, but will also provide thoughtful, nuanced insights for the broader industry. AI models are guided by the data they have access to, as well as the structures and inputs we create in building their infrastructure.

Yakov Livshits
Founder of the DevEducation project
A prolific businessman and investor, and the founder of several large companies in Israel, the USA and the UAE, Yakov’s corporation comprises over 2,000 employees all over the world. He graduated from the University of Oxford in the UK and Technion in Israel, before moving on to study complex systems science at NECSI in the USA. Yakov has a Masters in Software Development.

CM3leon possesses both the power and adaptability characteristic of autoregressive models, along with the remarkable efficiency and cost-effectiveness during both training and inference stages. This significant advancement overcomes the limitations of previous models, which were restricted to performing either text or image generation tasks exclusively. Most importantly, LLaMA won’t require researchers to have the large number of resources typically needed to train and run large language models.

In addition to the consumer-facing tools, executives at the meeting also announced a productivity assistant for employees called Metamate that can answer queries and perform tasks based on information gleaned from internal company systems. Earlier this year, Zuckerberg laid out his grand ambitions for generative AI in an internal meeting with company employees, speaking of the potential it has in creating 3D visuals for the metaverse. Meta is also planning to launch new generative AI features in Instagram that will enable users to edit images with prompts, plus “AI agents” for Messenger and WhatsApp, designed for education and entertainment. Meta, the parent company of Facebook and Instagram, launched a suite of generative artificial intelligence (AI) models on Aug. 2 called AudioCraft for music creation from various inputs, according to a blog post.

AI Sandbox is in limited release and will automate ad variant generation

This is the capability to generate novel concepts and creative variants faster and in higher volume than employing human creative professionals alone. Management jettisoned 10,000 employees in a drawn-out firing process that left the Menlo Park, California-based company without a tech road map and shook employee confidence in the direction for the business, sources have said. Meta chief executive Mark Zuckerberg told employees at the session on Thursday that advancements in generative AI in the last year had now made it possible for the company to build the technology “into every single one of our products”.

The comments came just two days after Senators sent a letter to Zuckerberg questioning the leak of Meta’s popular open-source large language model (LLM) LLaMA in March (which was seen by many experts as a threat to the open source AI community). Facebook parent Meta dipped its toe into generative AI this week by announcing AI Sandbox. Advertisers can use it to create alternative ad versions, use text prompts to generate backgrounds and crop images for Facebook or Instagram ads.

Earlier today, the company announced three generative AI features for advertisers. However, Thurai warned that Meta’s AI models won’t be able to follow the open-source path indefinitely. He said the company has to acquire large numbers of Nvidia GPUs, employ a wide AI talent pool and invest in other aspects of its data center infrastructure.

While Meta’s metaverse efforts haven’t panned out as expected, it still seems to be pushing on the idea of creating virtual worlds through generative AI. Bosworth told Nikkei that large language models (LLMs) — like OpenAI’s GPT-4 and Google’s PaLM — will help with 3D model creation as you’ll just have to describe them. Meta demonstrated AI Sandbox during a live media event in New York City on May 11 to showcase the technology as an arena for testing new generative AI-powered ad tools. The company plans to test out AI Sandbox with a select group of advertisers before opening up general availability in July. Generative AI, technology that creates text and images based on users prompts, has caught the attention of businesses across industries after the popular reaction to the wide release of OpenAI’s ChatGPT chatbot.

Snapchat released its AI persona, My AI, that allows users to prompt it for recipe suggestions, gift ideas, and content inspiration. AI teams from across the company will come together to focus on ways to implement generative AI for a more delightful experience. Find out more about the new top-level product group forming at Meta to develop generative AI experiences for Instagram, WhatsApp, and Messenger. Which begs the question, “Why not just build this into the main AI model and produce better results without this middle step?