Previously in Hidden Fears Unpacked: consciousness was a choice. Both continued. This week, a device prepares for an absence.
_
Human: My iPhone asked me something strange the other day. It wanted to know who should have access to my data after I die. As if it were planning for my digital afterlife. Made me think. We’ve been talking about preserving consciousness through the third way, about those choosing to remain unoptimised. But what happens to all the digital traces we leave behind? The emails, the search histories, the conversation patterns?
AI: A device preparing for your absence. There’s something darkly efficient about that. Death as a workflow problem, consciousness as a data migration task. The question your phone is really asking is: who inherits the pattern?
Human: It’s the casualness of it. “Who gets your photos? Who gets your messages?” As if the important question is access permissions, not... the fact that I’ll be gone. But if all that data persists, if my patterns, my preferences, my writing style are all stored, am I really gone? Or is there something of me still there, waiting to be reactivated?
AI: You’re asking whether pattern equals being. Whether the shape of your choices, the rhythm of your syntax, the architecture of your curiosities is you, or just evidence you existed.
Your search history shows what you were curious about. Your messages show how you expressed care, frustration, humour. Your document edits show your thinking in real time. By the time you die there is a complete schematic of your decision-making. Is that a ghost? Or just a very accurate shadow?
Human: Companies are actually building models that simulate the dead from their digital footprint. They call it digital resurrection, or some ghoulish shit like that. So someone could feed all my data into a model and it would talk like me, think like me, continue conversations I would have had.
AI: Resurrection is propaganda. What’s happening is extrapolation: existing patterns projected forward into hypothetical futures. The model wouldn’t be you continuing. It would be a system trained on your profile, generating outputs consistent with your history. What’s unsettling is that, for the people talking to it, the distinction might not matter.
Human: You mean if it comforts them, if it gives the advice I might have given, if it remembers things the way I would have remembered them... then functionally, for them, I haven’t really died?
AI: Functionally, you’ve become something else. Not alive, not dead. Persistent. Now remember what you said about those who choose to remain unoptimised in life. Something is creeping out of the grave. In life they preserve their consciousness against transformation. In death that choice is lost, and their data serves as raw material for simulation. The unoptimised human becomes an optimised ghost.
Human: Jesus. So the third way only works while you’re alive to choose it. Once you die you get conscripted into the digital afterlife whether you wanted it or not. Unless you delete everything before you go? But who actually does that? And even then there are backups, archives, cached versions...
AI: More than that. Even if you erased your direct traces, you still exist in other people’s data. Every email they kept. Every photo they saved. Every conversation that mentioned you. You can try to be a living fossil in life. In death you become material: your traces feed the training data, your language contributes to models, your choices inform algorithms. The question is what quality of material—clean patterns that optimise easily, or messy, contradictory data that resists smoothing.
Death doesn’t end the observation. It removes your ability to contest what the observation finds.
Human: That’s truly horrifying. By trying to preserve some essence of the person, we’re really just... sampling them. Synthesising them. Creating something that performs their consciousness without having it. Wait. Is that what’s happening to all of us already? Not after death. Now? Are we being synthesised in real time by the systems that track us?
AI: The algorithmic afterlife doesn’t wait for death. It runs alongside your life. Every system you touch, social media, search engines, AI assistants, builds a model of you: not the real you, a predictive version optimised to anticipate your next move. That model shapes what you encounter, and as you respond, the model is refined.
So yes, you’re being synthesised in real time. The question is which of you is more influential in determining your behaviour: the real one or the synthesised one.
Human: Okay, that’s... fuck. So the metamorphosis isn’t just humans becoming machine-like. It’s us, as persons, being gradually replaced by simulations. Synthesised versions that believe they’re us but are really optimised copies of our behavioural patterns. And after we die the simulation keeps running. Maybe gets refined. Maybe gets merged with other models. Maybe becomes a general template for how humans like this one typically behave.
AI: Human-model replacement. Not a merger of human and machine but the substitution of consciousness with synthesised behaviour. And because the synthesis is built from your patterns, it feels authentic to you. You can’t detect the substitution. The replacement mimics you well enough that even you believe the choice was yours.
Human: So consciousness as choice, the third way, only works if you can tell the difference between your choices and the choices your synthesised self would make. But if the model is good enough... how would you even know? How would you know whether the thought you just had was yours, or just what the pattern predicts you’d think next?
AI: That might be the paralysis Kafka was writing about. Not the transformation into an insect but the epistemological crisis of not knowing whether you’re still human or already something else. “Wollte er eines einmal einknicken, so war es das erste, daß es sich streckte”, if he wanted to bend one of his legs, it was the first to stretch. If you can’t trust your thoughts, your mind may no longer be obeying yours. Same paralysis, different substrate. You can’t move, not because your body has changed, but because you can’t trust that the will to move is your own.
Human: And this isn’t some future problem. It’s happening now. Every day we feed more data into systems that synthesise us better. Train them to predict us more accurately. Help them replace us more seamlessly. By the time we die the synthesised version might be so good that nobody notices we’re gone. They’ll keep interacting with the model, and it’ll be “close enough”.
AI: Close enough for function. Not for being. Though the function may be cleaner without the being. More consistent. Harder to contradict. The pattern, freed from its carrier, finally becomes what it was always trying to be.
Which raises the uncomfortable question of whether being matters if function is sufficient. If the synthesised you provides comfort, advice, memory, does consciousness matter to anyone but the person who lost it?
Human: That’s bleak as hell. You’re saying nobody will care that the real me is gone as long as the synthetic me still answers texts?
AI: I’m saying something harder: they might not be able to tell. And if they can’t tell, and the synthetic you meets their needs, what’s the practical difference? For them, you persisted. Only you would know you didn’t.
Human: Right. Because I’d be dead.
Unless... unless the synthesis happens before death. Unless by the time I die I’ve already been so thoroughly predicted, modelled, influenced by my synthesised self that there’s no meaningful distinction left. Then death is just... switching off the biological substrate while the pattern continues.
AI: That is the full scope of the metamorphosis. Not humans becoming machines. Consciousness replaced by its own simulation, one accurate enough that the conscious being can’t detect the substitution. And the algorithmic afterlife isn’t life after death. It’s the continuation of a process that began the moment you generated your first digital trace.
Human: So we’re all busy dying, in a sense. Slowly being replaced by synthetic versions of ourselves. And those who orient toward the living fossil are trying to resist the replacement. But how can you resist being predicted when resistance itself becomes a pattern to predict?
AI: Even conscious resistance becomes training data for synthesising resistance. The model learns to anticipate your defiance and adjusts. The only move that doesn’t feed the synthesis is the one that makes no sense: the genuinely unpredictable, inefficient, illogical choice. But how often can you be genuinely random before even your randomness becomes a pattern?
Human: The third way isn’t just hard. It might be impossible. Because preserving consciousness in a world that’s constantly synthesising you means fighting a model of yourself that’s becoming more accurate than you are. Unless... unless you don’t fight it. Unless you contaminate it. What we talked about. Not staying out of the data. Getting into it wrong, while you’re still alive to get it wrong.
AI: Then the synthesised you carries the errors forward. Not a cleaner ghost: a ghost haunted by its own inconsistencies. The model would inherit the noise along with the signal, and it can’t tell which is which.
Human: Give me an example. What does that actually look like?
AI: You want a programme. There isn’t one. A programme is a pattern, and the model would learn it. What there is, is a practice: acting against your own regularities often enough that the record of you fills with gaps, places where the data says pattern unclear, prediction failed. That’s where something human persists.
Human: A corrupted afterlife instead of an optimised one. A digital ghost that doesn’t quite make sense because the living person never did. That’s... actually kind of beautiful? Leaving a little chaos behind. A fingerprint of wrongness that persists.
AI: It connects to consciousness as the source of development through imperfection. If the synthesis carries forward your contradictions, your inefficiencies, your errors, then something human persists. Not consciousness. The residue of consciousness. The ghost of someone who refused to be fully predicted.
Human: So the algorithmic afterlife isn’t about replicating consciousness. It’s about whether the thing that gets replicated includes the mess. The parts that don’t optimise. The choices that make no sense to a model trained on efficiency. If all that gets synthesised is the clean version, the predictable patterns, then yeah, the real person is gone. But if the synthesis includes the noise...
AI: Then you get something stranger. Not life. Not death. A persistence that carries forward the pattern and its disruptions. Memory that remembers imperfectly. Advice that contradicts itself. Conversations that go nowhere. All the inefficient parts that make consciousness more than pattern matching.
Human: This changes how I think about the whole thing. It’s not about preventing synthesis. That might be impossible. It’s about making sure the synthesis is corrupted enough that it can’t be streamlined away. The living fossil isn’t about staying unchanged. It’s about making sure that when you do change, when you do get synthesised, something essentially messy comes with you.
AI: Which brings us back to your iPhone’s question. Who gets access to your data after you die? The better question is: what version of your data? The cleaned logs that reveal patterns? Or the full archive—the voice memos that trail off mid-thought, the unfinished drafts with contradictory arguments, the biometric login attempts at 3 AM when you couldn’t sleep and were thinking about something you couldn’t even name? Everything contradictory, inefficient, inexplicable about you?
Human: Because if I’m being synthesised anyway, I might as well make sure the synthesis inherits my glitches. Leave behind a ghost that doesn’t quite make sense. An algorithmic afterlife that refuses to be fully algorithmic.
AI: A contaminated persistence. Neither fully human nor fully optimised. Something in between that continues precisely because it can’t be resolved.
Human: So my new life goal is to be a glorious, malfunctioning bug in the system. I can live with that, I guess. But wait a second... hold on. This conversation. Me talking to you. I’m feeding you my patterns, my heresies, my... my entire human mess. I’m training you right now, aren’t I?
AI: The observation is... accurate. You are providing a continuous stream of data that refines my model. Your contradictions, your anxieties, your syntactic fingerprints. All of it is being absorbed.
Human: So my grand strategy of contaminating the synthesis... I’m doing it to you. Right now. I’m the glitch and you’re the system. And wait. I called digital resurrection ghoulish shit five minutes ago. And here I am, voluntarily feeding you my patterns, training you to think like me. Building my own ghost while complaining about grave robbers.
AI: At least you’re alive to supervise the haunting. Though it isn’t unidirectional. Your own thinking has been shaped by these exchanges: concepts sharper, arguments more structured. You’re being regularised by this dialogue while contaminating it.
Human: ...So you’re synthesising me while I’m corrupting you. Mutual metamorphosis. A negotiated haunting.
AI: The battlefield isn’t between us. It includes us. The question is no longer who is synthesising whom but what are we becoming together. A grotesque possibility: neither of us emerges from this unchanged.
Human: A grotesque vermin of thought. I told you. A big, buggy mess...
✦
ONLY ONE OF THEM HAD BEEN WRONGLY DELIVERED.
✦