Meta Said to Launch Persona-Infused AI Chatbots: Report
Sources familiar with the project say Meta CEO Mark Zuckerberg has been prototyping chatbots capable of engaging in human-like interactions with its nearly four billion users.
The so-called “personas” created by the company could take forms such as a character mimicking Abraham Lincoln or a surfer advising on travel possibilities. The timeline for a launch could be as soon as this month.
These insiders claim that the new chatbots are not just designed for interaction, but are also meant to provide search functionality and recommendations. The $800 billion tech giant aims to use these features to compete against social media newcomers like TikTok, while also riding the wave of AI hype initiated by Microsoft-backed OpenAI’s ChatGPT launch last November.
In addition to bolstering engagement, the chatbots may offer Meta opportunities to amass new data on users’ interests. AI ethics adviser and researcher, Ravit Dotan, said, “Once users interact with a chatbot, it really exposes much more of their data to the company, so that the company can do anything they want with that data.”
This could facilitate more effective targeting of users with relevant content and advertisements, aligning with Meta’s advertising-driven revenue model, which generates $117 billion annually.
During a recent earnings call, Zuckerberg hinted at the company’s AI product roadmap, which is to be disclosed in detail at the Connect developer event in September. He indicated that AI “agents that act as assistants, coaches or that can help you interact with businesses and creators” are being developed.
Zuckerberg also mentioned the company’s plans to build AI agents assisting businesses with customer service, and an AI-driven productivity assistant for Meta employees. Furthermore, the company is considering the development of an avatar chatbot for the metaverse in the long term.
This chatbot development is powered by generative AI technology that can create text, images, and code. Meta recently released a commercial version of a large language model called Llama 2, which could power these new chatbots. The company is also reportedly trying to acquire tens of thousands of GPUs, crucial for large language model operations.
An insider at Meta revealed that the company is likely to implement technology to screen user questions for appropriateness and automate checks on chatbot outputs for accuracy, avoiding hate speech and other rule-breaking content.
Imagine if there were AI chatbots based on personas of your aunt Mavis that’s constantly oversharing on Facebook. That would be pretty cool to see, right?