Meta creates AI version of Mark Zuckerberg so staff can talk to the boss

If you are one of Meta’s almost 79,000 employees and cannot get hold of the boss, do not worry. The owner of Facebook and Instagram is reportedly working on an AI version of Mark Zuckerberg who can answer all your queries.

The AI clone of Zuckerberg, Meta’s founder and chief executive, is being trained on his mannerisms and tone as well as his public statements and thoughts on company strategy.

The rationale behind the project, according to the Financial Times, is that employees could feel more connected to one of the most powerful people in Silicon Valley.

The Meta chief has a history of creating and experimenting with digitalised versions of himself.

In 2022, Zuckerberg shared his own avatar inside his self-proclaimed metaverse, which was publicly mocked for the graphic quality, leading him to post an upgraded version later. However, Meta has scaled back its vision for the metaverse, in which people’s digital representatives, or avatars, can interact with other virtual humans.

The company has instead been developing AI-generated 3D characters that can engage with humans in day-to-day conversation. It has recently started focusing on building a character based on Zuckerberg.

The 41-year-old executive, who is estimated to be worth more than $220bn (£164bn), is reportedly taking part in the process of training his animated AI. A person familiar with the project told the FT that the AI character would be developed using images and the voice of Zuckerberg. Meta believes the Zuckerberg experiment could be replicated by influencers and creators, a section of the digital economy that is wrestling with the notion of digital avatars.

Synthesia, a $4bn UK-based startup that makes realistic video avatars, said the idea of a senior company executive using AI to increase their internal presence was not science fiction any more.

“When you add realistic AI video and voice, engagement and retention go up significantly,” said a Synthesia spokesperson. “People work better when the information they need is delivered by a familiar face or voice.”

Until Zuckerberg launches his AI self, however, he will have to present in person at meetings with thousands of Meta staff, such as the one he carried out in 2023 two days after he announced that 10,000 employees would be laid off. Then, the tech chief was questioned by “rattled” staff about job security and the future of remote working.

The Wall Street Journal has reported that Zuckerberg could be helped to prepare for such sessions by a “CEO agent”, a personalised AI system that is being developed at Meta and is already helping him to get internal company information faster. Zuckerberg is driving Meta to use AI more internally, in the expectation that it will help lower costs and accelerate the pace of work.

Through integrating AI into its business, the company, which also owns the messaging service WhatsApp, aims to minimise its organisational structure and increase efficiency, which Zuckerberg has said is key to “get more done”. “We’re elevating individual contributors and flattening teams,” he said in January.

The reported moves form part of a company-wide effort to invest in AI in a drive to remain competitive with tech rivals that are also pouring billions of dollars into the technology. Zuckerberg is presiding over a multibillion dollar investment in AI in an attempt to create “superintelligence”, the term for a system that can perform any cognitive task far better than a human.

Last week, the company launched Muse Spark, an advanced AI model that it claims can estimate the calories in a meal from a photo and plan a family holiday by completing various tasks such as writing up a travel itinerary and searching for child-friendly activities simultaneously. The model has been praised for its performance in language and visual understanding but lags in coding and abstract reasoning.

Meta suffered legal setbacks last month when a jury in New Mexico ordered it to pay $375m in civil penalties for misleading consumers about the safety of its platforms and enabling harms, including child sexual exploitation. In the same week, a court in California found that Meta had deliberately made Instagram addictive and that a young user had got hooked, which led to her being harmed.

On Monday, Keir Starmer said social media platforms such as Instagram and TikTok needed to take action to stop young users wasting hours scrolling never-ending videos.

Britain, in common with other countries, is considering restricting access to social media for children and it is testing bans, curfews and app time limits.

“We’re consulting on whether there should be a ban for under 16s,” Starmer told BBC Radio. “But I think equally important, the addictive scrolling mechanisms are really problematic to my mind. They need to go.”

Meta has been approached for comment.