Engage with My Interactive Digital Avatar: A Personal AI Conversation Experience

Alexandru Voica, who leads corporate affairs at the innovative video-generation startup Synthesia, recently unveiled an impressive creation: an interactive virtual avatar designed to handle typical press inquiries about the company. This revelation followed a panel discussion where I had expressed my open-mindedness toward AI-generated content. Voica’s avatar represents, in my view, a significant evolution in the use of AI within public relations.

This past September, Synthesia invited me to their newly established office in New York City. Previously based in the U.K., the company stands out among digital avatar startups like D-ID, HeyGen, and Colossyan. Earlier this year, they achieved a remarkable $4 billion valuation and reported crossing $100 million in annual recurring revenue last year.

Synthesia enables companies to create interactive training videos featuring AI avatars. They have recently launched a product called Roleplay Sessions, which allows employees to rehearse tasks such as sales pitches with an AI avatar that reacts and evaluates their performance.

At the opening of the new office, I was offered the chance to have my own AI avatar, and without hesitation, I accepted. I was pleased with my appearance that day, making the prospect of a digital twin even more appealing.

Initially indifferent to avatars, I’ve since acknowledged their growing importance in our digital lives. Many users have started creating avatars on social media to enhance their content. My newfound interest led me to present my interactive digital twin, a first for Synthesia in crafting an avatar for anyone other than Voica. The avatar has been trained specifically to discuss my article on fraud in venture-backed startups and will only respond to inquiries related to that topic.

To engage with this feature, simply click “start in a new window.” You can pose questions such as:

• Why did I choose to write this story?
• What is the focus of the research paper?
• What findings did the researchers reveal?

The creation process involved a mini film studio within the Synthesia office, where multiple images were taken, and a two-minute audio recording was made of my voice. This required my consent, which ultimately led to the birth of “digital Dom.” My personalized avatar can read any script I provide and comes in versions both with and without glasses. Additionally, they developed two interactive versions capable of conversing with me.

The interactive avatar was trained on my selected piece, and a dedicated team constructed it. It employs a mix of voice-to-text, video, language, and text-to-voice technologies. The tech stack includes Synthesia‘s proprietary models, with options available for customers to use alternatives from other providers, including Cartesia, ElevenLabs, Google, or OpenAI. Businesses may also select their preferred cloud hosting service or pay Synthesia to facilitate the hosting.

The voice-to-text system converts spoken input into text, while the language model interprets and executes responses. The text-to-voice mechanism vocalizes these responses, and the video model animates the avatar during interactions.

Overall, Synthesia delivers three main products: a video-creation platform featuring standard avatars, an interactive platform named Sessions for avatar engagement in surveys or role-play scenarios, and an API platform allowing customers to amalgamate Synthesia’s models with various tech services for creating interactive avatars or other offerings.

Constructing my avatars took the Synthesia team just a few days. I first experimented with the standard ones and scripted a simple piece discussing the arrival of autumn in New York, my favorite season. The voice quality was impressively accurate, and it was free from any traces of hoarseness present during the recording.

Upon sharing it with some friends who are less tech-savvy, reactions were mixed, ranging from fascination to discomfort.

When I demonstrated the interactive version, which is designed to respond only to specific inquiries about my venture fraud article, I asked it where I worked before TechCrunch. Each time, it expertly redirected me to the article instead.

While my friends noticed differences in voice likeness and found the interactive avatar somewhat unsettling compared to the standard version, my mother found it “amazing,” a high compliment in her eyes. She and my father took turns trying to ask personal questions, but each time, the avatar directed them back to my article. “I don’t recall giving birth to two of you,” she quipped after testing the avatar.

This experience has profoundly influenced my perspective on the future of journalism. Would audiences accept news presentations delivered by an avatar? One investor dismissively answered no, highlighting ongoing skepticism surrounding the infiltration of AI into media. However, when discussing this topic with others, opinions varied. There are questions about whether avatars might enhance or potentially displace journalists. Would corporate leaders prefer interactions with AI representations over actual individuals?

In journalism, my passion lies in human connections, storytelling, and exploration. Yet, the essential element of trust remains pivotal—something that does not seem feasible to delegate to an AI.

Outside of journalism, the appeal of cloning oneself with an avatar is undeniable. It could eliminate the stress of catching up after a break, as a digital twin could answer inquiries in one’s absence.

As businesses explore avatar usage, my feelings about finally acquiring one are mixed. After the initial thrill had faded, I scrutinized my avatar closely, left waiting—for what, I’m uncertain. Perhaps a blink, a new phrase, or any form of acknowledgment that it understood.

Since my digital twins are deterministic, they will stay within their programming limits. However, I can envision how easily someone could develop a tendency for what some might term “AI psychosis” with a more dynamic avatar capable of unlimited discourse.

In discussions with an investor, I speculated that my generation, Gen Z, may struggle to embrace this technology. It seems to border on science fiction, with elements familiar from movies becoming a reality. Nonetheless, I find digital avatars less intimidating than humanoid robots; with an avatar, if the situation becomes uncomfortable, logging off remains an option.

For now, I present my personal avatar, summarizing this week’s top stories on our site!