As Mark Zuckerberg makes an AI version of himself… will your next boss be an avatar?

It’s obvious that Mark Zuckerberg, the founder and chief executive of Meta, the parent company of Instagram, Facebook and WhatsApp, would be a pioneer of emerging tech — no matter how dubious the rest of us might be. Perhaps most notably, he turned into a Sims-like avatar of himself and entered a virtual reality world as his company attempted desperately — and unsuccessfully — to make the metaverse a thing. Now the tech tycoon, estimated to be worth about £164 billion, is trying another way to reach his 80,000 employees: an AI version of himself, trained on data which includes his public and private thoughts on the company’s strategies.

The animated AI will also look like Zuckerberg and mimic his mannerisms, to help staff feel they have a personal connection to it. So, is Zuckerberg a trendsetter who is about to set off a revolution of always-accessible AI bosses, or is this, like the metaverse move, a flash in the pan?

“The first obvious reaction is that it’s a gimmick,” says Phoebe Moore, professor of management and the futures of work at the University of Essex — and a consequence of “over-inflated egos that drive the big tech stratosphere,” she adds. But beyond that, Moore thinks it’s a non-starter for a number of reasons. Key among them is how human employees interact with human managers. “Part of the whole point around how management relationships can work best is that you have an emotional connection,” she says. “You build each other up and you make each other better.”

But Jochen Menges, a professor of leadership at the University of Zurich and Cambridge Judge Business School, says Zuckerberg’s AI CEO idea addresses a real leadership problem: senior leaders “just don’t scale” and employees often lack access to the information, strategy and direction held at the top. “It struck me as potentially a really powerful solution, that this kind of AI CEO could close that gap,” he says.

By using an AI persona, Zuckerberg could in theory be in contact with Meta’s tens of thousands of employees — although that number is shrinking as the firm plans lay-offs of about 8,000 people in an efficiency drive.

Of course there are drawbacks: employees wouldn’t be interacting with Zuckerberg, but a digitised version of him. It seems unlikely major decisions would go ahead without the real person’s say-so. Anything different would pose huge accountability issues, reckons Lynda Gratton, professor of management practice at London Business School. “Who’s accountable for any decision?” she asks. “Because Mark hasn’t said it. His persona said it.” That’s something that Moore wonders about too, asking: “Who is liable if this AI gives the wrong answer somehow?”

Menges believes there are a limited number of cases where an AI version of a CEO could be useful, such as sharing consistent information or explaining strategy across a large organisation. But he warns that leadership is not merely a case of information transfer. Workers want to feel understood, see accountability and build trust with a real human being — all of which are difficult to simulate. “I really wonder if people want to be led by an AI CEO, or a duplicate of their CEO, as an AI agent,” he says.

The data on that is mixed: in a survey, 38 per cent of employees said they would prefer an AI boss to a human one, while half of those in the boardroom said they’d rather have AI managers than humans. Whether or not it works will soon be discovered by Meta’s employees — and its executives, including Zuckerberg. But Gratton believes if it’s anything other than a huge failure, it sets a precedent that others could follow. “It’s definitely a directional signal,” she says.