Databox CEO Peter Caputa lately posted that he’s releasing a brand new video course taught solely by his AI double. The avatar, powered by the favored AI video instrument HeyGen, seems and sounds identical to him, delivering hours of professional content material on his behalf.
This transfer has sparked a significant debate: Is that this an excellent time-saving technique or a step too far-off from authenticity? Whereas Caputa notes that he wrote each phrase of the script based mostly on his 25+ years of expertise, the individual on display is not him—it is a digital reproduction. And with HeyGen now valued at $500 million with over 40,000 enterprise clients, the query extra leaders will probably be compelled to reply quickly is:
Must you be making AI avatars of your executives?
To interrupt it down, I talked it by means of with SmarterX and Advertising and marketing AI Institute founder and CEO Paul Roetzer on Episode 167 of The Artificial Intelligence Show.
A Subjective Selection
Roetzer, who has identified Caputa for 18 years, has a nuanced take. As somebody who simply spent a whole lot of hours personally recording greater than 20 programs for SmarterX’s AI Academy, his instant response is that he couldn’t think about utilizing an avatar himself.
“For me, private connection and authenticity are important in speaking with my audiences,” he says. “I am unable to even fathom utilizing an AI avatar in my place to show a course.”
However, that does not imply he disagrees with Caputa’s selection, and the technique for him and his model.
“It’s a subjective resolution. There is not essentially a proper or unsuitable.”
Opinions on the topic are blended, too. Caputa’s LinkedIn submit on the subject drew tons of each unfavourable and constructive feedback. Some praised the effectivity of utilizing AI avatars, whereas others felt it created a disconnect.
The Uncanny Valley Is Disappearing, Quick
This query is about to grow to be far more urgent, says Roetzer. As a result of the technological limitations that when made AI avatars really feel clunky and unnatural are vanishing. The slight imperfections in motion or tone (that “uncanny valley” feeling) are being smoothed out with every new software program replace.
Quickly, telling the distinction between an actual video and an AI-generated one will probably be almost not possible with out entry to the metadata.
“We will get to the purpose the place you simply do not know,” Roetzer says. “Movies within the very close to future are going to only be indiscernible from actuality.”
That is not a distinct segment challenge. Roetzer factors to a current video of the US president that went viral on social media as a result of many viewers had been satisfied it was an AI avatar. Because the expertise turns into extra seamless, the road between actual and synthetic will solely get blurrier, forcing shoppers and creators alike to query what they’re seeing.
Discovering the Proper Place on the Human-to-Machine Scale
The talk over AI avatars mirrors a broader dialog about the place to attract the road with AI in content material creation. Particularly on condition that not all content material requires the identical stage of human contact.
Roetzer compares it to a scale he developed for writers utilizing AI throughout Advertising and marketing AI Institute’s AI for Writers Summit 2025. For some duties, like writing product descriptions or fundamental touchdown pages, there’s doubtless far much less debate over whether or not or to not use AI.
“Who cares?” he says. “Individuals simply need the knowledge.”
However for high-stakes, high-trust content material like a keynote presentation or a private editorial, it could get murkier, since audiences could count on the actual individual.
“When it is like a keynote presentation or an editorial piece, you need to know that that is coming from the individual,” says Roetzer.
The identical logic applies to AI avatars. Whereas utilizing one for a fast inner onboarding video could be acceptable, a paid course from a trusted professional carries completely different expectations. In the end, the choice comes all the way down to understanding your viewers.
“This isn’t prescriptive. It’s as much as you to resolve the place that consolation stage is,” says Roetzer. “In case your viewers expects you to indicate up and be authentically there, and to have put these additional two hours in to report the factor, you need to present up and do it.”
Each Chief Should Now Select a Path
Caputa’s experiment isn’t a one-off curiosity. It’s a preview of a selection each model and enterprise chief will quickly face. With avatar expertise turning into extra accessible and highly effective, the temptation to avoid wasting time and scale content material manufacturing will probably be immense.
The core query is not in regards to the expertise itself, however in regards to the technique behind it. It forces a dialog about what your model values most: the effectivity of automation or the irreplaceable connection of human presence.
For Roetzer and his group, the reply is evident: they received’t be utilizing AI avatars for his or her instructional programs . However he stresses that each group will need to have this dialog now.
“We’re all going to have to decide on how we use AI to create our content material, our thought management, our experience,” says Roetzer. “You will, as a model or as a person creator or chief, have the selection to deepfake the factor.”
