Synopsis
The Financial Times reports that he is personally involved in training and testing this digital character. The idea is that it could hold conversations with employees, offer feedback, and share thoughts in a way that reflects how Zuckerberg himself might speak.Meta has been developing photorealistic 3D AI characters that people can talk to in real time, the report said. The company is, however, currently prioritising an AI version of Zuckerberg himself.
The Financial Times reports that he is personally involved in training and testing this digital character. The idea is that it could hold conversations with employees, offer feedback, and share thoughts in a way that reflects how Zuckerberg himself might speak.
The system is being trained on a mix of Zuckerberg’s public statements, his tone of communication, and descriptions of his mannerisms. It also draws on his recent thinking about company strategy. The aim is to make interactions feel more personal, so employees feel a stronger connection to the founder, even when speaking to his AI avatar.
People familiar with the project told the Financial Times that this is still early-stage work. But if it succeeds inside Meta, it could be expanded further. In the future, influencers and creators might also be able to build AI versions of themselves, allowing fans to chat with a digital version of themselves.
However, the report stressed that this effort is separate from another AI system reported by The Wall Street Journal in March. That tool, described as a “CEO agent”, is designed to support Zuckerberg directly in his work. For instance, it helps him quickly find information and answers, reducing the need to go through multiple layers of staff to get updates or decisions.
This move follows Meta’s release of Muse Spark, the first large language model (LLM) from Meta Superintelligence Labs since Scale AI’s former CEO, Alexandr Wang, took charge of the artificial intelligence (AI) lab. It is a smaller AI model designed for tasks such as visual understanding and health-related reasoning.
Meta is also working on broader AI tools for users. One of these is “AI Studio”, which lets people create their own AI characters or even build AI versions of themselves to interact with followers and fans.
However, these developments have sparked concerns. Last year, reports suggested that some users created inappropriate or sexualised AI characters, which raised questions from regulators and child safety groups. In response, Meta restricted teenagers’ access to its AI character features earlier this year.
Despite the ambition, there are still technical challenges. The Financial Times report notes that making photorealistic AI characters work smoothly requires huge computing power. It is also difficult to keep conversations fast and natural without awkward delays or lag.
Meta has been improving voice technology for these AI characters. It acquired two voice AI companies, PlayAI and WaveForms, to this end.
