Mark Zuckerberg is building a photorealistic AI clone of himself so that Meta's 79,000 employees can interact with a digital version of the CEO instead of the actual one. According to the Financial Times, Meta is training the AI avatar on Zuckerberg's image, voice, mannerisms, tone, and public statements. The stated goal: so employees "might feel more connected to the founder through interactions with it."

Let that sink in. The solution to employees feeling disconnected from their CEO is not the CEO spending more time with employees. It is building a chatbot that looks like him. This is the corporate equivalent of a cardboard cutout at a book signing, except the cardboard cutout can talk and it cost hundreds of millions of dollars to build.

The project is separate from another initiative reported by The Wall Street Journal in March, where Zuckerberg is building an AI agent to help him complete CEO tasks. So there are now two AI Zuckerbergs: one for his employees to talk to, and one to actually do his job. At some point you have to ask what the real Zuckerberg is for.

In fairness, Zuckerberg is not fully checked out. He reportedly spends five to 10 hours per week coding on Meta's AI projects and participates in technical reviews. He is personally involved in training his own AI avatar. If the experiment works, Meta plans to let creators build their own AI clones. The company already showed off this concept in 2024 and briefly let Instagram creators make AI versions of themselves to reply to comments.

Here is the real story underneath the absurdity. Meta is testing the upper limit of what AI can replace in corporate hierarchy. If a CEO's public persona, tone, and mannerisms can be captured in a model, what exactly does face time with leadership provide that the AI cannot? And if the answer is "not much," that is a devastating indictment of how most corporate leadership actually works. The meetings, the all-hands, the town halls. If a chatbot can replicate the experience, maybe the experience was never that meaningful.

The timing matters too. Meta just spent $15 billion on a superintelligence lab. It burned $900 million in a single month on Claude API tokens. It is competing with OpenAI, Google, and Anthropic on frontier models. And the CEO's solution to a connection problem with his workforce is: build a robot version of myself so I can spend more time on the stuff that matters to me.

The internet's response has been predictable and correct. As one Reddit commenter put it: "The board should replace Mark with the AI clone and fire him." Another: "This would have been the easiest training of someone's mannerisms for AI ever. He already seems like he is AI." Brutal. Accurate. And exactly the kind of reaction that should concern Meta's HR team, because if your employees think a chatbot is interchangeable with the CEO, the CEO has a bigger problem than scheduling.