Nvidia: Avatars could replace human speakers

(Disclosure NVIDIA is a client of the author)

Nvidia this week presented its view of the future of the Metaverse at Siggraph. As expected, it was a multimedia showcase where each set of visuals reinforced the points various speakers were making. Much of the presentation focused on avatars and how Nvidia’s tools, by emulating facial muscles, could translate what might have been wooden performances by those avatars into a variety of visible emotions.

The end result was both more realistic and interesting. (I touched on this last week previewing Nvidia’s plans.)

What I found fascinating was that — in contrast to these avatar capabilities — the human presenters came across as wooden and unemotional. I’ve done a lot of presentations (and presentation training) over the years, spent several years acting, and used to be a competitive presenter. What we saw was undoubtedly a lack of prompter rehearsal; that often creates wooden performances for speakers who haven’t used a prompter in some time in front of a live audience. 

The contrast between the engaging avatars and the boring human speakers left me wondering why we need people for presentations anymore — especially since many of these presentations are now virtual.

The problem with people

I’m a natural introvert, which means I can get debilitating stage fright. At one time, I wanted to be an actor and it truly made that career path a no-go. I also attempted to become an attorney, which turned out to be another road not traveled. Then I watched a practice for the local college’s speech team. I tried it out on a lark with no audience and impressed the coach so much that he approved my travel to regional competition. I took first in my class, went to a state competition, won a medal, and ended up at the nationals a month later where I earned three bronze medals and a silver. My total competitive speaking experience was three months. 

I succeeded by putting aside my fears and focusing on emoting through the presentations, using those acting skills I had thought were a waste of time. 

Human presenters have issues. Like me, they can have stage fright, get sick, have personal issues that affect their performance, and they make mistakes. Once I was asked to step in and do a keynote for a CEO who was so inebriated at a dinner that he couldn’t stand up, let alone speak (not a great look for a CEO, by the way). Using the same approach I’d relied on for competition, I pulled it off. 

But what if you didn’t need to have humans speak anymore? What if you could have a replacement that would always be on, would have a perfect memory for the content, wouldn’t need a prompter, or notes, or cue cards?

What if you could build the perfect digital spokesperson?

Nvidia’s fix

At Siggraph, NVIDIA demonstrated it could take a photorealistic avatar, and have it automatically express emotions and vary speech cadence to transform a script into what appeared to be a live presentation. Instead of using a person, the company used an avatar that looked real but was, in fact, a digital construct. Issues like last-minute script changes would be far less of a problem; all you’d have to do is type in the changes, adjust the emotional parameters to match the words and suddenly you’d have a video representation that looked like the speaker had practiced for months. 

Nvidia argued that you could back up these avatars with AI capabilities that could allow them to change the speech content on the fly, to address questions, or even fend off hecklers in the audience. Imagine how this might be useful for political debate prep, where an avatar could be trained with the opposition’s talking points and past personal attacks and, using the opponent’s face, create a more realistic opponent than some staffer pulled in to do the job. 

The benefits of avatar speakers

Often, companies put employees on stage for the wrong reasons, usually revolving around visibility and face time. The goal of a presentation is to convey knowledge, not increase the visibility of a particular employee. By focusing on the wrong goal, we often end up with sad, wooden presentations from people who don’t like making presentations as part of their normal job. 

Avatars not only reduce stress on people who don’t like public speaking, but can improve the quality of a presentation because they can be optimized for the audience. Changing a few settings can mean a new accent, a different emphasis — even how the avatar is dressed.  

I expect that eventually we’ll see avatars take the place of live presenters since they can consistently outperform their human counterparts. And, at some point, when we get holograms working more realistically, we may see avatars take the place of real people at physical events, too.  At Siggraph this week, Nvidia showed that the age of human speakers may be coming to a close, and perhaps with it, the need for human actors and extras, too.