ILLUSTRATION BY GLENN HARVEY
Helen Todd posted some provocative headlines on her social media channels this summer: “Not Deepfaked!” “I Cloned Myself!” “Meet my Digital Twin!” They read like the plot of a science fiction movie, signaling a world where fantasy blurs with reality and awe mixes with anxiety.
In other words, the host of a Cincinnati-based podcast about creativity and artificial intelligence (AI) perfectly encapsulated the rhetoric swirling around the latest development in her career. Did you hear? Helen Todd has a clone.
On Instagram, Todd the human heralded the moment with a call for revelry: “Drumroll…meet my custom synthetic avatar: Helen 2.ODD” followed by a celebratory face emoji wearing a party hat and blowing a horn. “Yes, you read that right. I’ve digitally cloned myself, so now there are two of me: the human Helen and my custom synthetic avatar Helen 2.ODD.”
Host of the Creativity Squared digital platform, Todd finds the idea of self-duplication intriguing. So, a few days after the Instagram fanfare for Helen 2.ODD—pronounced “two point odd”—I meet up with her at her Finneytown home office to discuss the genesis and implications of her machine-generated twin and her writing and speaking on how creatives can collaborate with AI instead of fearing it. And she grapples with the big “why” questions: Why would someone want or need a digital clone, beyond the novelty angle? And why should we embrace the new technology?
Before we talk about the future, Todd takes me on a tour of her home on Winton Road. She’s repurposed the 1910 building, once used as offices for doctors and insurance agents. The previous owner’s renovations carefully preserved original features such as the hardwood floors and exposed beams, and Todd has decorated the walls with an eclectic array of paintings, prints, and art photos. She points out a series of canvas pieces representing the four seasons that were crafted by her mother using a modern take on 12th Century marbling techniques.
Todd grew up in eastern Tennessee without a television and doesn’t have one in this house. But the equipment needed for video and audio production—microphones, ring lights, electric cables—is everywhere. “This is where the magic happens,” Todd says as she gestures toward the green screen hanging on the dining room wall. She hasn’t taken it down since recording the video for the introduction of Helen 2.ODD earlier. It’s been a little hectic, she says.
We eventually settle in front of Todd’s office computer to trace the timeline of her synthetic avatar’s inception. Her face lights up as she scrolls through her social media channels, the place where she chronicles important life events. She points to a June Instagram photo of herself standing in front of a green screen in a production studio. “There will be two of me soon,” the post promises.
PHOTOGRAPHY COURTESY OF CREATIVITY SQUARED
The photo was taken at the Burlington, Vermont, studios of Render, a company that specializes in hyper-realistic avatar development. Engineers used 30 minutes of video and 12 minutes of audio Todd recorded that day to train artificial intelligence applications to clone her appearance and voice. A few weeks later, her RenderMe program was delivered, providing the means to convert words into videos of herself without ever turning on a camera. Helen Todd had been cloned.
She describes the feeling of embracing a disruptive technology in its infancy as both exhilarating and familiar. At age 26, she launched a social media marketing agency, Sociality Squared, in New York City. It was 2010, “before social media was even an industry,” she says.
Unveiling a digital clone in 2023 has reignited the thrill. But it’s also caused her to reflect on the unexpected directions new technology can take. “I loved social media at the time,” she says. “I still do, for better or worse, given what it’s become.” She pauses, searching for the right words. “I have a feeling going into AI that hopefully we’ve learned something from the social media experience.” Human Todd is an optimist.
Her career pivoted from social media marketer to artificial intelligence explorer in October 2022. A friend who works at OpenAI, which developed ChatGPT, showed her a demo of the program. She fed the AI app a rough scene from a miniseries project she’d been mulling for a couple of years, and it instantly produced a detailed script, complete with dialogue and director’s notes. “That moment captured my imagination,” Todd recalls.
OpenAI released Chat GPT to the public about a month later, and suddenly everyone was imagining the possibilities of generative artificial intelligence. And worrying about a future where machines could perform the roles of humans, eliminating jobs and threatening creativity.
Todd, influenced by her mother’s artistic background, was particularly intrigued by the impact of AI tools. She quickly decided to jump into the fray by creating a podcast exploring the intersection of art and artificial intelligence. From there, things moved quickly. She bought the domain for Creativity Squared on Valentine’s Day 2023. It wasn’t cheap, but Todd considered the financial investment an indication of her commitment. “I thought, OK, I’m doing it,” she says. Within a month, she developed a website landing page, a trailer, a YouTube channel and an e-mail newsletter. The podcast launched on April 20, and she’s put out weekly episodes ever since.
A friend connected her to Jon Tota, CEO and founder of Render, as a potential guest for the show. Todd had only a general awareness of cartoonish gaming avatars before meeting Tota, but she quickly saw the potential of custom synthetic digital clones and formed a partnership with the CEO. Render would become a podcast sponsor, and Todd would get a clone.
PHOTOGRAPH COURTESY OF CREATIVITY SQUARED
Making the arrangements for the cloning process took several months, though Todd began teasing her avatar’s arrival on her podcast almost immediately. She also frequently broached the subject in personal encounters with friends, family, acquaintances, and strangers. News of the pending hyper-realistic avatar drew a broad spectrum of responses—from “That’s so cool!” and “I want one” to looks of horror and disbelief. She remembers one conversation with a couple at a gallery exhibition featuring AI art. She told them about her soon-to-be clone, and they stared at her, mouths agape. “They were like, That’s terrifying!” she says.
Nicholas Caporusso finds Todd’s awe-and-anxiety experiences unsurprising. “Every single innovation ever invented has the same thing,” says Caporusso, a Northern Kentucky University computer science professor who researches human-computer interactions. Some see new technologies as an opportunity to advance humanity; others fear they will destroy it. He cites Plato’s 4th century B.C. dialogue, Phaedrus, as an example. In that case, the technology in question was writing and its potential to weaken people’s memories and spread false information. We’ve all seen how that’s played out.
In today’s world, Todd and her pioneering peers face a similar trajectory, Caporusso says. Her decision to clone herself places Todd at the forefront of the technology curve that starts with innovators, followed by early adopters and then the early majority. Synthetic avatars are still “very much in the innovators stage,” says Jill Schiefelbein, chief experience officer for Render.
Based on industry data from Render, Schiefelbein and Todd estimated in August that less than 1,000 known custom synthetic avatars existed worldwide. Todd speculates that her avatar might be the first of its kind in Cincinnati. But early adopters are coming within the next year, Schiefelbein predicts, and Todd and Render are helping accelerate the movement here by hosting an Avatar Experience at ADC Fine Art Gallery with opportunities for Cincinnatians to learn more about joining the digital doubles club.
As synthetic avatars multiply, Todd hopes Helen 2.ODD and her podcast can serve as vehicles to smooth the path for better human-machine relationships. “My mission is to envision a world where artists not only coexist with AI but thrive,” she says. “I want to be a proactive voice to shape the conversations around AI to be more human-centered.”
The most common uses for hyper-realistic avatars like Helen 2.ODD include personalized marketing, customer service, and sales materials, as well as educational and training videos. Todd sees her avatar as a way of producing video at scale. “We have to figure out what human Helen is good at and what Helen 2.ODD is good at,” she says. “I’m never going to film 25 individual videos welcoming guests to an event, but having my avatar do that would be a nice touch point for an event about AI.” Todd says she’s been reluctant to take on the additional workload of producing videos designed for the TikTok platform and is considering assigning that task to her clone as well.
Todd is sensitive to concerns about AI-generated avatars replacing human jobs. Her contract with Render gives her ownership rights to her digital likeness, an arrangement she sees as a crucial distinction from deepfake avatars created without seeking proper consent or providing fair compensation to humans. She also emphasizes the need for complete transparency when her avatar is performing. “I always want people to know when I’m using it,” she says.
But none of those steps will stop her from stoking the ongoing debate over how convincingly synthetic Helen impersonates her human counterpart. Even the people who know human Helen the best disagree.
Todd’s family members got their first look at Helen 2.ODD before a dinner at her brother’s White Oak home in late July. “I would say it was a little uncanny how similar the avatar was to my sister,” Chris Todd recalls.
A typical younger brother, he used the clone’s arrival as an opportunity to rib his sister about her own human imperfections. “One thing that I said jokingly, but also in seriousness, is that the avatar spoke almost too well for it to be Helen,” he says with a laugh. “Helen has a tendency to mix up little colloquial sayings and words. So I thought that was a bit funny.”
But after hearing a National Public Radio story about teachers interpreting flawless assignments as evidence students may have used AI to enhance their work, Chris thinks his joke might reflect deeper philosophical insights into the evolving relationship between man and machines. “Imperfections in speech are naturally human,” he says. “These avatars have to be taught imperfections.”
Too perfect or not, Helen 2.ODD drew rave reviews from most of Todd’s friends and relatives, including Chris’s wife and three children. “The kids were blown away,” he says. “They thought it was really neat.” And many people have told Helen that her clone’s voice is indistinguishable from her own.
Regardless, Chris Todd stands by his belief that he can easily distinguish his sister from her clone. He’ll concede that the distinction will likely be more difficult in three to five years. But, for now, something about the avatar’s mouth clues him into its computer origins. “If the mouth was a little bit sharper, that would go a long way to promote the realism,” he says.
On social media, Todd encourages followers to make comparisons as well. She edited the video with her initial Instagram announcement last August so that she and her clone appear side-by-side.
Visually, the effect is jaw dropping. The people appear identical—same curve of the smiles, same arch of the eyebrows. To be clear, Todd had intentionally enhanced the seeing-double effect. In a purposeful sequence of life imitating art, which in turn imitates life, she replicated her clothing and accessories for her human Helen video shoot to those she wore during her avatar recording session. “I wore everything the same except for my fingernails,” she says. (The clone wore red; human Helen went natural.) But the mirror-like impact of their common scoop-necked black T-shirt, black onyx ring, and silver infinity necklace is powerful.
At this point, though, Todd also acknowledges that her digital replica has limitations. The avatar displays elements of what artificial intelligence experts call the “uncanny valley” effect. “This is the space where our brains recognize that something is human-like but isn’t human-like enough to be human,” says NKU’s Caporusso, offering an example. “If we generate a person who has 18 fingers, that’s uncanny because we recognize that humans would not have 18 fingers.”
Caporusso describes the characteristics of machines and humans as two extremes. “Either we consider something to be purely synthetic like a robotic voice, or we have something that’s really human-like,” he says. “Anything in between is considered uncanny. There’s this feeling that something is off.”
In the case of Helen 2.ODD, the hands trigger my impression that something is out of whack. The avatar always keeps its hands clasped in front of its chest, occasionally moving them forward and back but never releasing the grip. This awkward mannerism, which human Helen duplicated for effect in the look-alike Instagram video, resulted from instructions during the creation process. The Render team advised clasping the hands to avoid the natural human tendency to flail our arms and make big gestures when talking. “There are certain limited movements the avatar can do,” says Todd. “It can’t do too many things with the hands.”
The digital avatar also has some limitations in expressions, a lesson Todd learned quickly when she sent her first avatar video to a friend. The friend had been a big supporter of her podcast but didn’t always listen to the episodes in their entirety. In the video, she prompted her digital counterpart to call her friend “the No. 1 fan of the podcast.” Then she hoped to add a touch of humor by sarcastically inserting the clause “although this may be disputed.” But the nuance of sarcasm was lost on the synthetic Helen, which delivered the line as a statement of fact.
Caporusso expects such obvious uncanny elements will soon disappear from avatars. “It’s just a matter of time,” he says. “The algorithms are quickly closing the gap with that uncanny valley and making content that’s so human-like it will be almost impossible to distinguish avatars from humans.”
Something about the confidence in his voice stirs a wave of anxiety in me, so I interrupt. “That’s so scary,” I say.
Caporusso shrugs and smiles. “It might be scary,” he says. “But it also might mean we could be golfing at the same time we’re giving a work presentation.”
PHOTOGRAPH COURTESY OF CREATIVITY SQUARED
Todd, like Caporusso, sees the progression toward more human-like AI avatars as unstoppable. But she downplays the idea of humans losing their identity in a world where machines can look and sound just like us.
Todd compares the experience of working with a digital clone to hearing a recording of her voice. “I don’t listen to it and think Oh, that’s me. It is me, but it’s not quite me in the way that I know me,” she says.
The human journalist asking the questions probes for deeper reflection, and Todd explains the relationship from a different perspective. “I don’t have an existential attachment to this clone,” she says. “I see it more like a digital puppet and not a replacement for who I am. It’s not Helen 2.0. There’s not going to be a Helen 2.0.” She chose the name Helen 2.ODD to play off of her last name as well as the strange reality of being cloned. Giving the avatar a name helps her think of it as a separate entity. “That kind of distances it,” she says. “It’s not me. It’s my digital avatar.”
Glancing back through her social media channels, Todd says that all art is representation. “As soon as it’s outside of ourselves, it’s representing,” she says. “When you put a photo on social media, is that you? Not at all.”
But even those comfortable on the innovation curve’s leading edge must honor the concerns of later adoptees, says Todd, who addresses this challenge during her introduction of Helen 2.ODD in Episode 14 of her podcast. “First and foremost, we support humans,” she says. “Our stance is that all AI tools should amplify human creativity and potential and never replace it.”
Todd has allocated a portion of her podcast’s revenues to ArtsWave, the nonprofit dedicated to supporting the arts across in Cincinnati. On Episode 14, she also notes that 60 percent of the show’s costs cover the people who work behind the scenes to make Creativity Squared possible. But she also uses the episode to showcase her synthetic avatar’s super-human capabilities. She’s programmed Helen 2.ODD to smile and declare “It’s nice to meet you!” in eight different languages. In contrast, she notes, her own greetings are restricted to English.
Back in her home office, Todd walks me through the process of generating a video of Helen 2.ODD. She types a prompt into her avatar app. Seconds later, she hits play and a voice that sounds a lot like the human I’ve been interviewing for the previous two hours recites the lines: Hi, Michele. Welcome to Creativity Squared studio. My human counterpart is excited to show me to you.
Todd then adjusts the settings to demonstrate the software’s features. She changes the speed so the avatar is racing through the lines. Too fast, we agree. Todd makes more adjustments. This time a slow, deep voice welcomes me to the studio. Todd shakes her head. “Yeah, I don’t like this one,” she says with a laugh. But her message is clear: Her synthetic avatar is a powerful tool, and in the right human hands it could be a boon for human communication.
“It’s a cloned voice and a cloned image,” she says, motioning toward the video screen of Helen 2.ODD expressing her delight at meeting me in Hindi, Portuguese, Polish, German, Italian, Spanish, and French, as well as English. “It’s synthetic media. It’s the most nonauthentic media out there, and I’m still wrapping my head around it too. But in some ways, it’s humanized communication in ways we can’t do. If I wanted to do a video message in Hindi and my intention is I’m trying to meet you where you are, I don’t think this is taking away the human part.”
She shakes her head. “I don’t think any digital tool will help replace the human connection, but can it help me crank out videos?” She laughs. “Yes.”
For the human Helen Todd, that’s cause for celebration. Bring on the smiling emojis wearing party hats.
Facebook Comments