In the not-too-distant future, I will live forever—digitally, that is. Thanks to the AI video startup Synthesia, my digital double now exists, and it’s eerily lifelike. Synthesia has been producing AI-generated avatars for several years, but their latest generation takes advantage of cutting-edge generative AI, resulting in avatars more realistic and expressive than anything I’ve ever seen.
The Data Collection Process
On an early April afternoon, I found myself at Synthesia’s stylish studio in East London. My mission? To become a digital clone. Tosin Oshinyemi, the company’s production lead, guided me through the data collection process. It felt like stepping into a news studio rather than a deepfake factory. The stylist ensured my outfit looked camera-ready, and the makeup artist tidied up my appearance. Hundreds of smiling Polaroids adorned the dressing room walls, capturing those who had been digitally cloned before me.
Synthetic Media vs. Deepfakes
Oshinyemi prefers the term “synthetic media” over “deepfake.” While both involve AI-generated videos or audio recordings of people saying or doing things that didn’t necessarily happen in real life, “deepfakes” have acquired a negative connotation. Think of unethical sexual content or political disinformation. Synthesia aims to create positive synthetic media applications, such as corporate communications and marketing materials.
The Promise and Ethical Questions
Synthesia’s avatars are impressive, but they raise significant questions. What happens when we can’t distinguish between real and synthetic? As technology advances, full-body avatars that walk and move around in spaces generated by individuals may become commonplace. The promise is exciting, but the ethical implications are profound. How do we navigate a world where reality blurs with the artificial?
Beyond Avatars: The Deepfake Landscape
Synthesia isn’t alone in this field. Microsoft’s VASA-1, for instance, generates hyper-realistic deepfake videos from a single image and audio file. While it opens up possibilities for education, communication, and entertainment, it also challenges privacy and security. Will facial recognition systems be bypassed? Can the biometric industry keep up with deepfake innovation? These questions remain unanswered as we venture further into the uncanny valley of synthetic media.
In the end, my digital double will exist—a testament to the convergence of AI, creativity, and our quest for immortality in the digital realm. But as we embrace these technological marvels, we must tread carefully, ensuring that our synthetic creations serve humanity rather than deceive it.
Comments