This new video AI model will disrupt many existing industries

A company called Infinity has recently released v2 of their AI model, which can generate AI avatars with eerily realistic expressions and emotion.

After you finish reading this post, you'll be able to create videos like these in a few seconds.

AI Avatars?

Many AI startups have tried to reaching this AI avatar milestone — HeyGen, and Synthesia to name a couple — but without much success.

In case you are not aware, most people use the term "AI avatar" to refer to a realistic, human-like character who can also speak in a realistic AI, human-like voice.

To make an AI avatar, all you need to do is:

1. Choose a face.

Say this guy from HeyGen:

2. Paste the text you want your AI avatar to speak.

In the background, the AI tool will generate a video of the character saying whatever you want it to.

The problem with existing AI avatar software

They look fake. And robotic. With no emotion.

I mean, just go to Synthesia and look at some of their avatars to get an idea.

Some companies have tried to overcome this problem by using pre-recorded videos of people saying something. The software then modifies their lip movements to match your text. ArcAds is one such tool that does this.

The problem: There are some words that are, let's say, not appropriate for certain emotions. For example, if you're a lawyer, you don't want a person to be overjoyed when they say the word "divorce".

You get the idea.

The next iteration: Expressional AI avatars

Infinity AI has just come very close to solving this problem.

To get things started, you need an image, text and voila! you get your results.

Here are some examples from real users.

The comments on Hacker News about this tech are mostly positive (a rare thing for the HN community), so we're definitely onto something.

This technology has the potential to disrupt industries

Here are some of them:

Faceless YouTube videos

There is a growing trend of people trying to create YouTube videos without showing their face.

The problem: Millions of years of evolution. We're simply attached to people.

Technology like this enables those YouTubers to create videos with an actual avatar face behind it.

Currently, this model has a 30 second limit, which is about to change very soon, hopefully.

AI short video ads

These video are ~30 seconds long. This type of technology makes those videos look more appealing.

Virtual influencers

Once this tech gets a bit better, expect realistic videos from your favorite influencers about all sorts of topics.

E-learning

Imagine an entire course made with an AI avatar.

Doing that currently is hard because these AI avatars are...boring. We want to keep people engaged, and having an engaging face is a huge part of that.