September 11, 2026
India is moving corporate bonds onto tokenized rails connected to its digital rupee. We are taking systems that were already complicated and giving them AI, automation and eventually quantum computing. Nobody knows what anything costs now so just wait until we have a thousand AI agents inventing and deploying financial derivatives nobody understands around things nobody understands. But none of that was the story that really got me thinking. That came at the end of today's show, when I ran across a guy named Kun Chen who decided to distill himself into an AI.
Chen had a practical problem. His AI agents kept coming back to him whenever they did not know what decision to make. So instead of continuing to answer the same kinds of questions, he started encoding his judgment. He built files describing his opinions, the tools he uses, how he approaches problems, how he researches things and even how he writes. In other words, he started taking pieces of Kun Chen and turning them into markdown files. Then an AI could load those files and make decisions that were hopefully a little more like the decisions Kun himself would make. He even published the framework. That part is interesting enough by itself.
Then he did it to other people.
Chen built AI versions of Sam Altman, Elon Musk, Dario Amodei and Mark Zuckerberg using their public interviews, testimony, blog posts, social media and other material. Then he stuck the four artificial personalities into a room and had them argue about AI. And yes, they argued. That is amusing, but I honestly do not care very much about what the fake Elon Musk said to the fake Sam Altman. I would never take one of these bots, show it a car I was thinking about buying and assume its answer was genuinely what Elon Musk would tell me. That is not the interesting part. The interesting part is that Chen could build the thing at all.
Think about how much of yourself you leave behind in the things you create. Suppose you are 50 years old and you have been writing since high school. You have essays, emails, articles, social media posts, recordings, interviews, maybe decades of professional work. Whether you meant to or not, you are in that material. Your values are in there. Your sense of humor is in there. Your preferences are in there. The way you solve problems is in there. How you respond when somebody disagrees with you is in there. Humans shape the things we make because we cannot help it. If enough of those artifacts survive, AI can begin working backward from the artifacts toward the person who created them.
It will not reconstruct your soul. I am not suggesting that for one second. When somebody dies, whatever you believe happens to that person afterward is a completely separate question. But the artifacts remain, and those artifacts may contain enough information to reconstruct some recognizable portion of a personality. Right now I think of these AI personality bots like a tiny thumbnail image. If the original image is 1,400 by 1,400 pixels and somebody gives you a 128-by-128 thumbnail, you can recognize what you are looking at, but blow it up and everything gets fuzzy fast. That is probably where we are with personality reconstruction. Low resolution. But technology has this irritating habit of increasing resolution.
So what happens at 4K?
That is where this gets ethically weird. What happens when somebody clones you without asking? What if the reconstruction becomes good enough that your family says, “Yeah, that sounds exactly like him”? What if somebody takes that version of you and makes it do things you would find disgusting? Chen put four AI executives into a debate. Fine. What if somebody puts high-resolution personality reconstructions into a virtual cage match and tells them to kill each other? They are still software. I understand that. But they are software deliberately constructed to reproduce identifiable real human beings. Somewhere between a crude chatbot wearing an Elon Musk name tag and an extraordinarily detailed reconstruction of Elon Musk, there is probably a line. I have no idea where that line is, and I am pretty sure nobody else does either.
Then consider the opposite use. Your husband or wife dies and you are nearing the end of your own life. You have thirty years of their emails, photographs, home videos, voice recordings and messages. Maybe you would rather spend your last six months talking to an imperfect reconstruction of that person than sitting alone. Is that wrong? I do not know. It could be enormously comforting. It could also be psychologically disastrous. And who gets to decide? Does the dead person have to consent before death? Does the surviving spouse get to decide? Can the children build DadBot because they miss him? Can a company build one because his public material happens to be available on the internet?
Those are not science-fiction questions anymore. The current versions may be fuzzy thumbnails, but the basic distillation process already exists. And that is what bothers me more than four billionaire chatbots fighting about artificial intelligence. We are getting very good at building the technology before we have even finished identifying the questions the technology creates. The real issue is no longer whether pieces of a human personality can be extracted from the artifacts that person leaves behind. Apparently they can. The question is how high the resolution can get before that reconstruction becomes something we have a moral obligation to take seriously.
Circle P
SoapMiner
Tallow Soap
https://soapminer.com/
Discount code: Use "BITCOINAND" in the Cart's Coupon Code box for 10% off total purchase
AI personality cloning
digital clone artificial intelligence
AI personality reconstruction
Kun Chen AI distillation
ethics of AI digital humans
Member discussion: