Meta Patents AI to Mimic Deceased Users Online
Meta has been granted a patent for an artificial intelligence system designed to simulate social media users, even after they have passed away. The technology, detailed in a patent titled “Simulation of a user of a social networking system using a language model,” could allow an AI to continue posting content, interacting with others, and maintaining a digital presence as if the user were still alive.
The patent explains that a language model can be trained on data generated from a user’s past actions on the social networking platform. This trained model can then simulate the user’s behavior when they are absent, whether due to a long break or, more significantly, death. The stated rationale behind this technology, as outlined in the patent, is to prevent the user experience of others from being negatively impacted by a user’s permanent absence from the platform. The idea is that the digital presence of a deceased individual could continue to engage with their network, thereby maintaining the platform’s overall engagement and content flow.
Echoes of Past AI Ventures
This development by Meta is not the company’s first foray into AI personalities. Last year, Meta launched AI chatbots that were designed to embody various personas, including those of celebrities like Tom Brady and Naomi Osaka. These AI personalities were intended to offer users new ways to interact on the platform. However, the initiative reportedly struggled, with users finding the alternative personalities “weird and creepy.” Meta eventually discontinued the project, highlighting a potential disconnect between the company’s AI ambitions and user expectations for genuine human connection on social media.
The failure of these celebrity AI personas underscores a key challenge: social media’s core value lies in human connection. “Bots don’t fit that value,” the transcript suggests, emphasizing that users seek interaction with real people. While AI can be entertaining or useful, it often lacks the authenticity and nuanced understanding that defines meaningful human interaction. An AI trained solely on a user’s posting history might replicate word choices and patterns but would likely fail to capture the true essence of a person’s personality, especially for meaningful engagement.
The “Project Lazarus” Speculation
Adding another layer to the narrative, an anonymous post circulated online, purportedly from someone within Meta, detailing a project codenamed “Project Lazarus.” This alleged internal communication described an AI capable of taking over a deceased person’s social media, creating age-progressed photos, interacting with content, and essentially continuing their digital life. The post, which predated Meta’s patent filing, painted a picture of an AI that was highly capable of impersonation, suggesting it could seamlessly manage multiple accounts even if an entire group of people disappeared. While this information is unverified and originated from a 4chan post, it aligns with the implications of Meta’s patented technology and has fueled speculation about the company’s long-term AI development goals.
Broader Industry Trends and Ethical Concerns
Meta is not alone in exploring the concept of digital immortality. Microsoft filed a similar patent in 2021 for an AI chatbot that could simulate deceased individuals, fictional characters, or celebrities. This patent included plans for recreating people through photos and videos, potentially enabling 2D and 3D digital representations. While not explicitly stating a “bringing back the dead” function, it used deceased loved ones as an example of how the technology could be applied to imitate real-world individuals. This suggests a broader industry recognition of the potential for AI to simulate human presence beyond physical life.
Potential Use Cases and Societal Impact
Mark Zuckerberg himself has previously discussed the possibility of using the metaverse to bring people back virtually, particularly in the context of helping individuals cope with grief. In an interview with Lex Fridman, Zuckerberg acknowledged that while such interactions could be helpful for reliving memories, there’s also a risk of becoming unhealthy. He suggested that the primary use case might involve interacting with historical figures or individuals with unique knowledge. However, the potential for AI to act as digital clones of deceased loved ones raises significant ethical questions.
Grief and Digital Immortality
Sociologists and ethicists express concerns about the impact of such technologies on the grieving process. “One of the tasks of grief is to actually face the loss,” notes a sociology professor from the University of Virginia. Engaging with an AI simulation of a deceased loved one might complicate or hinder the natural process of accepting loss, honoring their memory, and moving forward in a healthy way. The ability for an AI to continue posting, liking, and commenting on behalf of a deceased person could fundamentally alter how society understands and experiences death and remembrance.
Business Incentives and User Adoption
From a business perspective, the incentives are clear: increased engagement, more content, and valuable data for training future AI models. For influencers or businesses, an AI clone could theoretically maintain an online presence, respond to messages, and engage with content even when the real person is unavailable. However, the prospect of interacting with AI versions of deceased loved ones or even living celebrities is met with widespread apprehension. The transcript notes that the “vast majority of people will not want this,” viewing it as dystopian and “very, very weird.” Marketing such a service without significant backlash would be a considerable challenge.
Legal and Technical Hurdles
Beyond societal acceptance, legal challenges loom. In the United States, 23 states recognize postmortem rights, protecting a deceased person’s identity for up to 100 years from unauthorized commercial use. Companies would need to navigate these complex legal landscapes, potentially through user agreements that allow for the waiver of such rights. Technically, creating a truly convincing AI simulation that captures a person’s full personality, including their nuanced communication and emotional responses, remains a significant hurdle. Even advanced language models are primarily pattern-matching systems, and replicating the depth of human connection is an immense challenge.
The Unpredictable Future
While Meta has stated it has no current plans to implement this specific patent, the granted patent itself, along with similar filings from Microsoft, indicates a clear interest in this domain. The rapid advancement of AI technology, including real-time audio and video simulation, suggests that such capabilities may become more feasible and sophisticated in the coming years. The question remains not if, but how, these technologies will be integrated into our digital lives, and whether users will embrace or reject the idea of digital ghosts lingering on social media platforms.
Source: Meta's New AI Is Freaking Everyone Out… (YouTube)
