Chapter 1: The Intersection of AI and Personal Identity

In recent months, I've been experimenting with various generative AI platforms that transform photographs into videos and synthesize my voice using AI-generated text. My research for my Ph.D. centers around crafting a digital representation of myself through a blend of portrait photography and video technology.

While I continue to seek the optimal combination of AI tools to accurately reflect my identity, I'm mindful that I may not be sharing the most pertinent aspects of myself. After all, I'm fortunate enough to not be facing the end of my life just yet. Now, let’s delve into the platforms I’ve explored, complete with images and links to the resulting videos and audio.

As I engaged with ChatGPT to draft a script for the AI audio tool, I posed several thought-provoking questions:

• What insights might artist Ginger Liu offer regarding AI's potential to replace traditional family portraits?
• Can AI truly recreate a person's essence, and how might this affect the grieving process?
• Will AI serve as a substitute for portrait photography as a marker of identity?
• Is it feasible for AI to capture a person's true essence more effectively than a photograph?

ChatGPT provided some compelling points which I refined for clarity.

Section 1.1: The Art of AI Portraits

The topic of identity and essence in generative AI portraits is particularly captivating. As an artist, I believe that a person's essence transcends mere physical appearance; it encompasses the intangible qualities that define their character and spirit. Although generative AI can create images that closely resemble the subject, it falls short in capturing the deeper essence that a human photographer might convey.

Despite this, I don’t view generative AI portraits as inherently negative or unethical. They can offer valuable artistic explorations and prompt important discussions about identity and authenticity in our digital landscape. However, it’s crucial to recognize their limitations and not confuse them with genuine human representation.

Subsection 1.1.1: The Role of Photography in Memory

As an artist, my curiosity drives me to explore innovative forms of art and technology. I am currently involved in several projects that integrate AI, eager to witness how this field will develop in the future. Ultimately, I assert that the most impactful art emerges from a profound understanding of human experiences—something that AI cannot replicate.

I hold that photography possesses a unique capacity to encapsulate and immortalize memories of our loved ones. With the advent of AI, there has been a growing trend in using generative AI portraits to simulate the presence of those we've lost. This phenomenon raises significant questions regarding the nature of essence and identity in AI-generated works compared to traditional photography.

Although AI portraits might achieve a realistic likeness, I believe they cannot authentically replicate a person's essence or the memories and emotions tied to traditional photography. In this regard, AI may never truly replace photography’s role in helping us navigate grief.

However, I caution that AI could potentially extend the grieving process if not approached thoughtfully. The ethical ramifications of using AI to simulate interactions with departed loved ones or generate related content must be carefully considered. We must be mindful of the possible effects on those experiencing loss and ensure that AI is employed in ways that honor the dignity and memory of those we have lost.

Chapter 2: Exploring AI-Generated Audio and Video

In my journey, I utilized ElevenLabs to recreate my voice. I began by recording a minute of my speech and uploading it to their platform. After that, I added the text generated by ChatGPT, and within moments, I received an accurate reproduction of my voice.

The next step involved using D-ID’s digital human creator. I uploaded a high-quality still from a recent interview and paired it with a casual photo taken from my phone. After that, I integrated the ElevenLabs audio of me reciting the ChatGPT script.

Despite my expectations, the results were surprising. Below are the videos that illustrate my experience.

Final Thoughts

While ElevenLabs is still in beta, it managed to produce an impressive clone of my voice, which is remarkable; however, it also raises concerns about the potential misuse of such technology. D-ID's attempt to replicate my speech movements was somewhat lacking, though it excelled in mimicking eye blinks and head movements.

Both platforms proved to be user-friendly and could be valuable tools for businesses seeking affordable video solutions for their websites and social media.

Special thanks to Ethan Mollick for his insightful contributions.

Ginger Liu is the founder of Ginger Media & Entertainment, a Ph.D. researcher specializing in artificial intelligence and visual arts media, as well as an author, journalist, artist, and filmmaker. Listen to the Podcast.

Watch Video