Could we soon send an identical “AI twin” of ourselves to meetings instead of our cat? According to Zoom CEO, yes.

Eric Yuan, CEO of Zoom, gave The Verge an in-depth interview to explain his company’s AI vision. Like many organizations today, Zoom wants to AI-first Companies, even if that’s not the quote we remember from the interview. According to Yuan, in a few years we will have a “digital twin” of ourselves attending meetings.

“Suppose the team is waiting for a decision from the CEO or maybe a meaningful conversation, then my digital twin can really represent me and also be part of the decision-making process. We’re not there yet, but that’s one reason why there are limitations in current LLMs,” says Yuan.

An LLM for everyone

The introduction of generative AI into the context of digital meetings is not new. Zoom has also AI Companion This helps you take notes or remember the gist of a conversation. Yuan wants to go a step further and replace your physical presence with AI, although this is not yet possible with current technology.

“Everyone is using the same LLMs now. This doesn’t make sense: I should have my own LLM. Essentially, this is the foundation for the digital twin. Then I can count on my digital twin. Sometimes I want to participate, so I join. If I don’t want to participate, I can send a digital twin to participate. This is the future,” Yuan said during the interview. According to the CEO, this could be a reality in five to six years.

Deep fakes

There are indeed some reservations about Zoom’s vision. First, it puts pressure on the integrity of digital interactions. If someone can create a realistic deepfake of themselves, you will never be sure who the person on the other side of the screen really is. This can pose serious risks: a few months ago, an employee of a multinational corporation transferred millions of dollars to a fake account after being deceived by a deepfake of his boss.

Another issue is hallucinations. How can you monitor what your AI twin is doing or saying during a meeting? You don’t want to get in trouble because your digital copy suddenly goes completely off the rails and your boss scolds you, to name just one example. According to Yuan, these are problems we will be solving in the coming years.

Experts from the AI ​​world think differently. At X and its competitor Bluesky, critics dismiss Yuan as an out-of-control tech CEO. “The idea that an LLM can predict a person’s response seems completely wrong to me,” writes researcher Simon Willisson on X.