Meta Challenges Tech Rivals by Announcing AI Chatbots With 'Personalities' Across All Platforms

Meta Platforms CEO Zuckerberg's move into the generative AI space is an attempt to reclaim users that have drifted to rival social media apps such as Snapchat and TikTok. Challenging its competitors like Google, Amazon and Microsoft, the company plans to create AIs with distinctive personalities, interests and views.

According to Zuckerberg, it is not going to be just about "answering questions" but about "entertainment and helping you do things to connect with the people around you".

Meta revealed the introduction of Meta AI in beta, which is an advanced "conversational assistant", available on WhatsApp, Messenger and Instagram, and is also going to come to Ray-Ban Meta smart glasses and Quest 3.

Furthermore, Meta AI is said to offer users "real-time information and generate photorealistic images from your text prompts in seconds to share with friends". However, it is available in the US only.

Meta is launching 28 more artificial characters, now in beta, with unique interests and personalities. These will include cultural icons and influencers such as Snoop Dogg, playing a "Dungeon Master", Korean chef Roy Choi giving you cooking tips as "Max" or Kendall Jenner playing "Billie" the role of a big sister.

The feature can also enable you to talk to tennis player Naomi Osaka as "Tamika" about manga or chat with MMA fighter Israel Adesanya as "Luiz" about MMA.

So essentially you can replace your Facebook friends with AI friends.

The focus of generative AI is on the creation of current and original content, synthetic data, chat responses and deepfakes. It is useful for novel problem-solving by autonomously generating many types of new outputs. For instance, chatbots and virtual assistants are based on generative AI with its reliance on Natural Language Processing. ChatGPT, Bard, Alexa, Siri and Google Assistant are such generative AI tools.

The main difference between generative AI and traditional AI is in what their objectives are and how they function. Whereas traditional AI is based on using predefined rules and patterns to perform tasks, generative AI exceeds this limitation by creating new data similar to human-created content.

Moreover, to differentiate between its rivals Meta will use its extensive social platforms to place AI in front of its users. The firm stated that it does not believe in "one single super-intelligent AI that everyone uses", but relatively "different AIs for different things".

Meta AI will have a new assistant that will allow users to "interact with like a person" during group discussions or to help resolve family arguments, for example. Although this resembles ChatGPT and other programmes, the company also announced the launch of an AI studio, where people and developers will be able to customise their own AIs.

The tech giant further explained the reason behind creating an AI studio is to allow users to "create your own AI that's aligned with your goals, whether you're a small business, a creator or anyone really". Essentially, any assistants created within the AI studio will be accessible to users within any of Meta's messaging platforms. The company also said that the various chatbots would have their own profiles on its platforms for user interaction.

Zuckerberg also said that although as yet, the celebrity chatbots are responsive in text only, voice responses are probable by early next year.

Additionally, Instagram will launch two new AI image-editing features: Restyle and Backdrop. Restyle will allow users to apply different visual styles based on a text description. For instance, "watercolour" or "collage from magazines and newspapers, torn edges". On the other hand, Backdrop changes the scene or background of your image based on prompts.

Speaking about these Instagram AI tools, Meta stated: "We know how important transparency is when it comes to the content AI generates, so images created with restyle and backdrop will indicate the use of AI to reduce the chances of people mistaking them for human-generated content. We're also experimenting with forms of visible and invisible markers."