Mark Zuckerberg, the CEO of Meta, has unveiled a new AI-powered assistant, Meta AI, designed to integrate seamlessly with Meta’s ecosystem, including Instagram, Facebook, WhatsApp, and the company’s mixed reality devices. This move puts Meta in direct competition with OpenAI’s ChatGPT.
At the recent Meta Connect event on September 27, Zuckerberg introduced Meta AI, powered by the company’s substantial language model, Llama 2. The development of this AI assistant involved collaboration with Microsoft Bing, allowing users to access real-time internet information effortlessly.
Zuckerberg described Meta AI as a “basic assistant that you can talk to like a person.” Beyond answering questions and engaging in conversations, this AI can also generate images using a new tool called Emu. Meta trained Emu on an extensive dataset comprising 1.1 billion pieces of data, including user-shared photos and captions from Facebook and Instagram.
What sets Meta AI apart from its competitor ChatGPT, according to Zuckerberg, is its approach. Instead of creating a one-size-fits-all chatbot, Meta aims to develop various AI products tailored to specific use cases. To illustrate this point, Zuckerberg demonstrated how Meta AI could be customized for each of Meta’s social media platforms. For instance, it could assist in coordinating travel plans in Facebook Messenger group chats.
Meta’s chatbots aren’t just about providing useful information; they are designed to be conversational and entertaining, enhancing the user experience.
In addition to the announcement, Meta revealed a collection of chatbots based on around 30 celebrities, including Paris Hilton, Snoop Dogg, and Tom Brady, with an entertainment focus.
Meta AI is set to become available on September 27 for a select group of users in the United States on Facebook Messenger, Instagram, and WhatsApp.
Furthermore, Meta AI will be accessible to users of Meta’s upcoming smart glasses, scheduled for release on October 17 in the U.S., as well as the new Quest 3 VR device.
On the same day as Meta’s Connect event, OpenAI made an announcement regarding ChatGPT. The chatbot will no longer be limited to data from before 2021. This update, available immediately for Plus and Enterprise users using the GPT-4 model, addresses the knowledge gap that had been widening due to ChatGPT’s prior limitations.
The previous knowledge cutoff in 2021, tied to the AI model’s training data, has now been extended, allowing ChatGPT to provide more up-to-date information and insights.