"Kendall Jenner", Deepfakes, and The Identity Economy

for when life starts feeling like a Black Mirror episode

Kendall Jenner’s Meta AI persona “Billie”

Anthropomorphized chatbots are facets of generative AI that simultaneously evoke feelings of deep fascination and skepticism in me. Fascination because like any new technology, they pay testament to the limitless power of human innovation. Skepticism, on the other hand, arises from the potential dangers of anthropomorphization through the emergence of a data black box attributed to their ability to access personalized data.

About two months ago, Meta launched an AI character chatbot, “Billie” modeled after Kendall Jenner. The idea being put forth by Meta and the language used around Billie is that she is your “big sister” whom you can go to for wisdom and counsel. The most chilling aspect of this all was the video unleashed where Billie (Kendall Jenner’s deepfake) is seen addressing her audience and reassuring them that she is there to chat and that they should message her for any advice. I have spoken about anthropomorphization before when discussing the effect of customization on perceived closeness with AI through exploring Snapchat’s My AI. Meta has taken it a step further by taking advantage of the parasocial nature of relationships that form the foundation of celebrity and influencer culture. Kendall Jenner (Billie) is not the only celebrity but rather one of about a dozen who have been paid by Meta to grant them permission to use their image.

public figures who are part of Meta’s AI campaign

I confess I have never tried this chatbot and have no intentions to do so on the grounds of Meta’s past violations of data privacy laws. As a singular data point, my use or lack thereof of Meta’s AI chatbots is insignificant. I am but a drop of water in an ocean. However, I would like my resistance to plant the seed of skepticism. I encourage users to question why a corporation known for violating data privacy laws is willing to pay celebrities a hefty sum to use their identity to stipulate friendship with an AI chatbot. I came up with a list of insightful questions that allow us to take a deeper-dive into this matter:
What information in the form of personal data am I likely to provide to a friend (read: anthropomorphized chatbot) that cannot easily be extracted using cookies and other forms of observed data?
If these chatbots are trained iteratively using Natural-Language Processing, what is the benefit of encouraging a large number of users to have many human-like conversations with them?
What is the expected return on investment (ROI) when a for-profit company like Meta is willing to pay celebrities millions of dollars for only six hours of work in a studio?

These queries got me thinking about the role that identity plays in Generative AI. I first came across the term “Identity Economy” while reading a blog post written by my dear friend and futurist Sylvia Galluser. The term in itself is a form of hidden genius based on my own interpretation. If we look at economics as “the study of how we make choices in the face of scarcity and how those choices motivate behavior”, then we can look at the economy as the playing field where scarcity, choice, and behavior interact. Identity on the other hand is who or what something or someone is. Put together, the identity economy explores choices and behavior in the face of identity “scarcity”. Scarcity here refers to the relational value of someone’s personal identity i.e. how famous someone is and how much influence they can exert. In this way, reflective of the world we live in, we all have identities but in the face of generative AI and the identity economy, some identities have higher value based on the amount of influence someone has.

Sylvia’s blog post provides a chronological flow of events from a futurist perspective; from the inception of ChatGPT to the centrality of public figures in AI safety. Focusing specifically on the latter, there have already been cases of public figures such as Andrew Huberman and Cardi B having their identity stolen for marketing purposes. As deepfake systems increase in both sophistication and availability, these cases of identity theft will inevitably rise. Deepfake technology can be used to manipulate the masses by disseminating propaganda and engaging in unethical product marketing. In such situations, there are drawbacks for each stakeholder. The general public seeks to lose out by spending their hard-earned coin on products they believe are endorsed by celebrities. Celebrities and public figures, on the other hand, could have their public images tarnished if their identity is used to market controversial ideologies, products, or services. Businesses could also lose out in deepfake attacks through defamation by fradusters who seek to damage their reputation.

On the flip side, the proliferation of deepfake technology creates a great opportunity for accelerated progress in data and identity protection. Ideally, the prospect of identity theft should drive influential figures to become advocates for robust regulations. Since celebrities have a plethora of resources and substantial influence,their engagement in addressing the challenges posed by deepfakes may catalyze collaborative efforts between industry leaders, policymakers, and technology experts.

While the development of deepfake technology is captivating, it carries with it an abundance of potential risks. When addressing risk, policy, and societal change it is imperative to understand who holds the reins of decision-making power and who seeks to lose out the most. The intersection of technology, identity, and influence demands thoughtful consideration, as we attempt to chart a course towards a future where innovation exists within ethical guardrails.