The Afterlives of Our Digital Selves
What becomes of us after we’re gone when AI agents and our digital selves continue to be active online?
A good friend of mine from when I was younger passed away back in 2016. But people have continued posting on his Facebook profile and in a remembrance group over the years. Admittedly, every now and then, I’ll take a look at his profile—it’s almost like just seeing something he created on the internet gives him some solidity in my world.
Similarly, one of my earliest friends on Facebook who I had never personally met or knew outside of that social platform also passed away long ago. She died in a car accident in 2011, and yet I just now logged onto Facebook and I see comments from as recently as this past November saying things like “I remember you”, “Miss you”, lots of Happy Birthday messages on her birthday each year, and one comment from 2022 where the person wrote “A great spirit visited all of us so briefly… but with a lasting memory.” Her death in 2011 inspired me to write about my own thoughts on our digital selves and what becomes of our internet presence after we die.
Some data from ExpressVPN suggests that by 2050 there may be some 186.9 million profiles on Facebook from people who have died. There are similarly millions of other accounts on other social platforms from people who have passed, with the number of profiles of deceased people growing more and more.
Most of us on social media have come upon those instances of the profiles of no longer living people having some activity for some time after their deaths, slowly facing the inevitable creep of inactivity.
But something new is now happening for our digital selves online.
With the growth of AI (and what I think will soon be an explosion of AI agents that we all use to accomplish daily tasks for us), it’s becoming more and more likely that our social profiles and our digital selves on the internet will take on a life of their own, acting on our behalf (and, hopefully, acting as we would want them to) to various degrees. While some of us may wish to retain as much authority over our digital selves as possible, I can envision a near future where some people have recreated themselves online as agential AI avatars who autonomously interact for them on social media and various websites. Heck, maybe some of us will choose to have multiple versions of ourselves, maybe some of them even “appearing” to be rather different than who we are in the real world.
What happens if we die and our autonomous digital selves continue functioning? Could we literally have afterlives in the internet experienced by other versions ourselves?
Echoes of the Self
As I wrote in Echoes of the Self: Consciousness and Identity in SOMA, Star Trek, and Altered Carbon:
Is it possible that we might one day understand human consciousness well enough that we could take whatever it means to be you in conscious awareness and then put that consciousness into a computer, a robot body, an organic body, a virtual reality, or some other realm? Can the consciousness be separated from the body?
We can also explore what might happen to our digital versions of ourselves—whether sentient or not—who act on our behalf online, interacting with our friends and family, managing our calendars, representing us at online meetings, and even doing things like trading, posting, commenting, shopping, selling, and more.
We certainly don’t know enough about consciousness yet to know if brain uploading or transferring our consciousness to new bodies or to digital realms is possible. But we should easily be able to see that it’s only a matter of time before we have our own AI avatars who we can choose to give access to various levels of autonomy on our behalf.
The proliferation of AI technology has ushered in a remarkable new era where many people are now using LLMs like ChatGPT and Claude to do their homework, conduct research, help them write, create recipes, sculpt fitness plans, serve as first-steps for medical advice, and more (I’m personally doing all of these things myself). Just the other day I used ChatGPT to help me hunt down a scene from a TV show that I knew little about but I had seen a short clip in an Instagram reel some time ago—together we found the show and the scene so I could watch it on YouTube. And this morning I created a new project in ChatGPT for exploring various facets of historical mythology.
The ways that we’re all using AI (and we pretty much are all using AI, whether we like it or not, if we’re doing just about anything online these days) varies greatly from person-to-person. Some people aren’t actively interacting with LLMs by choice (and might not even be aware that almost all customer support platforms and even many of the emails and messages they see online are already being generated by AI). But some other people are going far more into the realm of offboarding their work and even their thinking to AI. Some people have even taken to interacting with AI chatbots like they are friends, therapists, or even lovers.
And the realm of AI agents is on the rise! AI agents are already being employed across a range of industries and processes. Agents not only have the processes of machine learning and language processing that we see in LLMs, but they are given access to autonomously conduct operations on behalf of people. Pretty soon many of us will have our own AI agents who can schedule our meetings, speak with us so they know how to answer our text messages, will help us reroute using our map apps when we come upon traffic on the highway, they’ll schedule medical appointments on our behalf, they’ll scour the internet to find out where we should stop for the cheapest gas and groceries nearby, they’ll have conversations with us when we need to mull over an idea, they’ll interact with our smart home systems to alter the temperature and airflow of our homes before our arrival, they’ll monitor our home camera systems to let us know if there is a potential intruder we need to be aware of… they’ll take anything we’re doing with LLMs and add in the capability of acting on our behalf online.
And that means that we will likely soon see our social media profiles acting as AI agents on our behalf. Some people will avoid this at all costs. Some won’t want to interact with profiles being driven by AI. But, for some people, the potential for having your profile always active and always generating content on your behalf will be too alluring. Some AI agents could turn people into social media influencers early on (before the tech advances such that all AI supported profiles are capable of driving maximum engagements from human and AI profiles alike). I personally can imagine a near future internet where the version of me on each social platform is supported with an AI agent that scours those platforms looking for content that I want to view myself or that it should respond to on my behalf (and the crazy thing about this possible future is that we can train these systems to respond exactly as we would respond).
If you can imagine this future as well, then perhaps you might be wondering, as I am, what would be to happen to such AI agential social profiles if we should physically die.
Could there be AI versions of us online in perpetuity?
Some social platforms are already devising strategies for how to deal with profiles of people who have died. But these have had their own issues. For instance, after Elon Musk took over Twitter and renamed it X, the company announced they would remove any profiles that had not had any activity for several years. This led to a lot of backlash as some of those unused accounts could be from jailed political dissidents or from oppressed peoples, but they also can be accounts of people who have died but who have family and friends who still want to see their accounts on social platforms.
As our possible futures unfold, including one where social media is awash in AI driven profiles that never sleep, are always active, and are developed to replicate our behaviors on those platforms, then the social media companies themselves may need to develop better strategies for handling not only AI profile behaviors but also for addressing the requests of the real humans involved. We might have the option (or be forced) to set our profiles up such that if we die our AI profiles will be able to learn that and deactivate themselves and/or create memorials for us. There might be some route where family or friends can report that a person has passed away (though that could also be problematic if some reporting system is abused—as is always the case with social media reporting systems it seems).
But what if there is no way to effectively stop the proliferation of AI profiles in social media from people who have died? It’s already expected that the number of profiles of deceased people across Facebook, Instagram, TikTok, and X will outnumber the entire U.S. population before the year 2075. What if the future of social media isn’t just to have one or several AI agents acting on our behalf on our various platforms, but also to have versions of us “living” and acting on those platforms long after we die?
I have no idea what social media may look like if that future should come to pass, but I’d bet it’s going to be pretty weird.
AI Integration and Our Digital Futures
The integration of LLMs and AI agents into various facets of life suggests the need for a reevaluation of our digital identities and what it means to be “online.”
As this technology grows ever more autonomous, AI agents will begin to act on our behalf, making decisions, interacting with others, and even forming relationships. If we do develop AI driven social media profiles, then it might even be possible that some AI profiles will make relationships with others, both human and AI, that the profiles will choose to cultivate even without our immediate knowledge. And those AI versions of us might live on after we are gone.
The era of AI is advancing rapidly, and the blurring of lines between human and machine challenges our very notions of selfhood and agency in the digital realm. How much of ourselves and our own authority over how we are viewed by others will we choose to share with digital versions of ourselves that run autonomously? Even while many people think that we are nearing a time when our medicine and technologies allow us to reach “longevity escape velocity” and to practically live forever, we might find that the breadth of who we are is replicated by digital versions of ourselves who can live well beyond us in the digital realm. Heck, I can even imagine a future where I set my AI avatars on social media and the internet to act on my behalf while I escape into the wilderness for weeks at a time to think and live without any connection to the digital realm for my physical self.
One more bizarre thought I’ll leave you with here: what if there is a future where many of us have online versions of ourselves running and operating on our behalf and then something should happen that wipes out the human species entirely (nuclear warfare, biological weapons, alien invasion, etc.), leaving behind a digital record of us stored on physical media. And, just what if, some alien intelligence finds that record and turns it on within a digital realm of their own? In such a case, perhaps our first contact would be from the digital versions of us meeting aliens in a virtual space.
The potential to store ourselves in digital form is also intriguing for sending messages to space. We may one day send out not just records of who we are and what we know in messages, but we might send digital versions of ourselves (perhaps even active within some stored virtual realm) to travel through space and time to take us to distant worlds.
Honestly, I would love to see my deceased friends again in the digital realm. Even if I knew it wasn’t really them, but AI agents acting on their behalf and trained on everything they ever shared online, it would still feel like at least an echo of who those people were. And, as the tech gets better, it might even be nearly impossible to discern between the echo and the real thing. We may first reach AI integration with our lives online long before it should happen through any physical alterations.