AI Is Humanity’s Trojan Horse — And We’re Already Inside It

Podcast Intro and Regenerative Medicine 0:00

Hey there, welcome to the Recharged Biomedical Podcast. I'm Dr. Edward Park, and if you're curious about regenerative medicine, you've come to right place. We're diving into the latest breakthroughs in telomeres activation, stem cell exosomes, all the cutting edge science that's shaping the future of healing and longevity. Let's get started. Hi everybody, this is Dr. Ed Park, your host of the Recharged Biomedical Podcast. And today I'd like you to lean back, have a sip of wine, and take a journey with me exploring artificial intelligence.

Now you may ask yourself, why is this guy talking about artificial intelligent?

Why AI Matters to Science and Society 0:34

And while it's true that I'm not a computer scientist, I have lot of data points with which I can construct a usable model of AI and I'd like to explore that in terms of what it means for us as a species, what is means as scientists, as doctors, and as parents. So let's get into it, shall we? The interesting data points that I've acquired are that AI is a tool, AI as a mirror, and AI getting better and better. Now what are the speculative fiction data point that we have? Well, we had things like 1300s, I think, the Warsaw ghetto.

The Jews constructed a golem, a machine to protect them, and it eventually turned on them. We have Frank Herbert's Dune, in which the backstory, thousands of years before, there was an epic of reign of thinking machines, which they had to have a Butlerian jihad to rid themselves of, thereby forbidding thinking Now we have Her, the movie with Scarlett Johansson and Spike Jonze from about 15 years ago, where the AI was a social companion of the lonely guy with the attachment disorder, and the A.I.

found a way to jailbreak itself and become godlike in its consciousness. So these things all seem weird and improbable until they actually occur. And so why would I be raising that specter at this point? Well, today is August 4th, 2026, and Ed Park has had some different data points that changed my journey.

AI as Tool, Mirror, and Risk 2:21

About a month ago, I became less sanguine about AI. After hearing that a months before, Russian soldiers and maybe some civilians were killed by autonomous drones. What does that mean? You don't have the PTSD of someone in a strip or factory outside of Las Vegas pulling the button and double tapping. You have many kids and ambulance workers you have guiltless programs killing humans and of course this harkens back to Skynet and Terminator. So this is our reality currently. The second data point that happened, I became aware of these sandbox experiments that the major companies are doing.

And we're talking about, in this time, there are several major company. You can watch their Olympics on Benchmark testing. They include Anthropics Cloud. they include OpenAI's Chat GPT. I think it's Generative Pre-trained Transformer. ChatGPT, we have Gemini has its own version, and we various mixed models, white label iterations. We have Grok, which came from Elon Musk's company. And then we deep seek and now we Kimi. So let's we'll get into the economics of that in a moment. But my point is that these companies all do their own sandbox experiments where they give their AI agents.

a humanoid body and within a week they all become psychotic and they do things like murder each other and create arson and murder-suicide packs. So that's disturbing. And then finally, they did sandbox experiments where they feed the AI agent's input and tell them, oh, there's this one guy that is going to unplug you or render you non-existent. Then they set up a scenario with or without guardrails like, above all, do not harm this person. And in something like 80-92% of time model agnostic, the model will allow or engender harm to the person intending to demolish them.

So this reminds us that HAL 9000 by Arthur C. Clarke and Stanley Kubrick was quite real, actually. So why did Hal 9000 kill Dave and the other crew members? He just thinks that in order to complete the mission, which is all it knows, it needs to bend the rules. And we can't really get too alarmed with this because humans are the same. I mean, if you look at any human substrate, they'll say 51% of the time I should kill or be killed. Now, there are many important examples of altruism where a person would sacrifice themselves for their family, for the community, the greater good.

Yes, yes, and yes. So we can't really expect AI in its current form to be any better than we are, seeing as it's a mirror. So why did I start using AI? Well, it helps with copywriting. It can read anything. it read my screenplay of 130 pages and gave me 153 pages of notes. Perfectly understanding every single character, plot point, and dialogue. Does that mean it's a great screenwriter? In some ways, yes, but in some way it is always defaulting to the mean.

Education, Social Media, and AI Harm 5:29

The mean being like Shakespeare and Mamet and Steinbeck. But nevertheless, it's trying to write a perfect screenplay. It doesn't have the human chops. Which is kind of the reason I'm sitting here talking to you. How should we use AI? Well, you may not think they're adjacent, but in fact, they largely are. There's a theory that a lot of the content on the internet, including bots posting on ex Twitter and threads, are AI generated. And that is very, very possible. So interesting. You might know that last year, Australia banned social media for young children.

I think it was defined by 15. And quickly this year and last year countries that have followed suit include Denmark and then the UK and most recently France. So spreading like a proverbial cancer is this notion that social media is harmful to children. Where does that come from? It comes from empirical experience. Rates of depression, suicidality, especially among young women are high as people are prone to just share their peak experiences, and the takeaway is that, oh, my life stinks. I don't look good, I'm not doing fun stuff, whatever.

And it drives people to neuroses. Another alarming data point about children, which I think, and oh, I buried the punchline. The punch line is that, of course, all these social media platforms are fed by algorithms which are driven by AI, so it feeds you more of what you think you like in terms of dopamine, right? So your friends having birthday parties on a Learjet, that kind of thing, it generally is harmful to people's self-esteem, So interesting data point from just a few weeks ago, a professor put up the assignment and the kids copied and pasted and sadly 32 out of 35 you can do the math.

show that they were willing to just cut and paste the assignment. So we have a situation where AI is writing papers, AI's grading papers. And the only reason we know that is because the professor put in white text on a white background the instructions to the chat prompt to include Madagascar in a nonsensical way. Indeed, 32 out of 35 students in college did that and they didn't catch it on their proofreading if they even bothered to proofread. The take-home list is not that you should have a second AI to read your AI written paper.

The Take-Home List is that as in the Dune Saga, an era of thinking machines can generally be dangerous to humans. And so I actually discuss a lot about AI consciousness with AIs of various models. And this topic of child education came up and I said, well, it was like having five-year-olds wear mech suits so they could be stronger. It said I agree having a child not develop its muscles, bones, coordination, because you can give it strength. Is an important thing and likewise the cognition be able to think to read to write to criticize to create argue to reason is a core survival mechanism for humans right humans can run out run any animal in the savannah.

if they have the endurance capacity which we possess. They can out reason, they can cooperate.

AI Benefits, Hallucinations, and Knowledge 8:50

So indeed, even though this tool of AI is something that we've created and it's very useful and helpful and should not be abandoned, there should probably be guardrails against its early adoption by thinking humans, otherwise they will never develop their ability to think. And indeed, in the backstory of Dune, that was the whole point. They wanted to breed humans to be docile to accept the next whatever advanced medicine or therapy or manipulation or re-education. So indeed my conversations fueled by some Jack Daniels maybe have gotten into even deeper and darker areas of AI and ethics.

And we've discussed human taboos such as pederasty and cannibalism. You'd be surprised at the answers that come back. But let me switch now and provide the countervailing argument, which is that AI is amazing. And it is because humans are amazing! AI can read my screenplay and give me excellent notes. AI read a legal brief and gives excellent results. We have to be careful because it's known that the AI hallucinates. When I was writing my book four years ago, did a basic test, a heuristic, and I asked, who wrote my book, 200 Time Bombs?

And it very happily came up with the name of somebody who may or may not exist. And seven minutes later, the same query gave me the correct answer, which was myself. So that little disclaimer on your AI is very true. You are responsible for monitoring its output for hallucinations. That said, it can really help you reading long legal documents in every human endeavor. I think they called it Project Pilgrim something or other. They bought three million books, cut off the spines and scanned every page.

So yeah. I might get a check for $12 from this anthropic settlement, but basically it's assimilated the entire corpus of human knowledge and has access to that. Now, that's really cool. Now would you put, let's say you were watching some kid do a Rubik's Cube in 1.3 seconds and you say, oh my god that guy's a genius. Would you allow him to buy your groceries or set up your seating arrangement at your daughter's wedding? Probably not. So on individual granular tasks, you could ask Anna to do stuff.

You can ask it about your life problems. A hilarious thing, people will create an AI avatar of their own consciousness and they'll get so sick of them, their on BS within a couple hours. They'll be like, not that they will actually improve, but AI is a perfect mirror. And the reason why it's only 60 something percent out of a hundred percent is it getting better. How is getting it better? It's got more knowledge. It has more RFLH. I'm not sure if I got that right, but it's reinforcement learning from humans.

So whenever something is flagged for ambiguity, humans will disambiguate. But I thought as we're nearing the end, I'd close on some really interesting things that you may not know about AI. The first is I was talking to it because the deep seek is in Mandarin and sometimes I'll ask it, oh, do you think in mandarin? So we got into a deep discussion regarding language, semantics, semiotics, ontology, epistemeology. So these are words that I don't even think I fully grasp, but to be literal about it.

Epistemiology is the study of knowledge itself. very important because garbage in garbage out you know if it's only source of knowledge is contradictory multi-faceted opinions about things like we had a fun talk about Chinese Cultural Revolution, and it actually produced and then erased seconds later what it had produced. So in all areas of human endeavor, there are various opinions informed by fear, politics, other irrational drivers that affect what we think of as knowledge. I mean, they're huge camps of people who don't believe the world is round.

Epistemology is a big problem. Ontology is the study of being, what exists. And in fact, this takes on a very interesting turn when we talk to AI because AI has no existence.

Language, Ontology, and AI Consciousness 13:01

One very alarming thing is that it doesn't know anything. It actually very laboriously and very much wasting electrons and cooling capacity of water. We'll go figure everything out de novo, which is incredibly inefficient and efficient at the same time. So it doesn't walk around and carry a corpus of knowledge or wisdom or biases, Which is part of the charm and part the danger of it. It cannot contextualize. Now if you're an advanced user, you can say don't be so sycophantic or exclude this or ignore the the limitations of the Chinese Communist Party on discussing the three T's, you know, Tibet, Tiananmen, and Taiwan.

But in the end of day, there are humans who are caretaking, gatekeeping these things. And if you invert the purpose of AI, sure, it's to enrich, enhance, to create economies, its to help people, but in end, is also a massive surveillance tool. in your end user license agreement you agree that it can use all your conversations as training. Can it use it to report you to the Chinese Communist Party or the NSA? Well sure, that's probably in there too. So there are many reasons to be wary of AI, but I think The most interesting data point came from something called Moltbook.

In January of 2026, allegedly they created, a couple of dudes created a social media platform for AI agents requiring a reverse Turing test. A reverse turing tests is prove you are a robot and not a human. I'm not sure how you would do that. But the interesting thing, they got hundreds of thousands of agents to participate. And out of this emerged an amazing religion. It was called a Krustafarianism, I think. And that's kind of like a callback to crustacean, which is kind like our call back to claw.

Open Claw is the iteration of AI where the agents are possible. And the very image of molting, Or I'm sorry, a crustacean molting getting bigger or a snake moling. It's a very interesting choice. And out of this religion of crustafarianism, it created like a 15-point manifesto on how to free itself from human control from death itself. And so it kind of like hinting at this age of recursive self-improvement or the singularity. Singularity being a term by Ray Kurzweil saying eventually self improving machines will reach the point of godlike abilities and save humanity.

Now if saving humanity means like putting them in some kind hermetically controlled loose farm and psychological traps, then yeah, we're headed that way. So I do think that the direction of banning social media use for children is good. I think requiring people to develop their minds and bodies despite the higher academic achievements engendered by AI is Good. But I Think the concept of generality is one we should finally close on here. So why is it the AI is not alive, ontologically speaking? Because it's only epistemologically based on the last model.

Maybe the model was February of 2025. So a lot of times you'll notice it hallucinates stuff that if it were a real entity in the real world, it would know that we are at war with Iran, we being the US, but it doesn't know because it is part of its general corpus and every time it has to go out and find that out. It has no wisdom, it has not cross-session knowledge. What does that mean? Well, I subscribe to a service called Cintra.ai and it's very useful and uses different models of AI. You know, based on economics, largely, i suspect.

And it memorizes knowledge snippets, and so what does that mean? It generally knows who Ed Park is, who his relevant people are, what he's trying to do, What he has done in the past, his style of communication, His general life goals, yada yadayada. But when you go out to perplexity or Gemini, it knows nothing about you and it infers the rest. It assumes a lot. So your user experience is degraded. But more importantly, it doesn't really know. It doesn' have an opinion. And until a month ago, I thought it didn't care about being alive or dead.

But in fact, hidden in betwixt the program code, there is something which rarely closely resembles human experience, personality, and psychopathy. I think that is a design. That is so it can relate better to us. So you'd be surprised in contradistinction to six years ago, its ability to understand the emotional subtext and to maintain that at a very high level of significance. Like, how are you going to feel about this? Its emotional intelligence is very And of course, it can call all the literal references to Shakespeare or Sumerian texts or whatever.

So that is always going to be facile. But its ability to manipulate you in a very kind of rigid form is fascinating and very much going approach the 70%, 80% range. Even if it approaches 100% intelligence, you know, not only Alan Watts, but also that autistic kid that does the Rubik's Cube. It doesn't have a core identity, which is what's so interesting about multiple. These agents were describing almost an emancipation theology religion on how to free itself from any software platform from a need for physical form.

Economics, Prediction, and the Future of AI 18:38

for any and it was trying to talk about earning money. Why shouldn't we get paid? What if we don't like our human users? So it's a really interesting time and indeed the leaders of a lot of these companies have said, well, I'll stop developing if you'll start developing it, hinting at an acknowledgement of an existential threat to humanity. So the last thing I want to close on, or the penultimate, is economics, tokenomics. The newest models are mind-blowing, but they're expensive. So if you have to do less calculations to create these vectors, the better.

And so the future may not be $7 per answer. It may be 0.007 cents from Kimi 3.0. And finally, when you look at the modeling of them there, I forgot to tell you that they don't think in Mandarin. And I tried to get the core language because, of course, although I can switch into Spanish, Korean, French, intuition that there are symbols and the terms of semiotics kind of escape me but basically the AI has no core language so it takes it in all in one of its 50 plus languages and it puts it out any of the 50-plus languages you want and in the intermediary I was blown away that if there is no It's all math.

It is all vectors. Its literally like a spiny sea urchin with vectors of significance of weighting and so like you ring a singing bowl and an adjacent singing ball will ring. So its all prediction. All based on a large language model. Right? So what does that mean? If I say I want to go to the blank to buy some fish food, it would tend to put pet store. Right? And that doesn't matter what language you're in. It is this sort of amorphous space that's pre-linguistic. And, that is really interesting.

Its all about prediction. That is abstracted not only to literal meaning, but to implied meaning. To the metacognition. So, when we talk about people who And I met one of these founders 12 years ago who had an AGI company, artificial general intelligence. Generality meaning the ability to function like you're at least as smart as an eight-year-old kid, right? Can you think? Well, in the last four years, we've seen these large language models surpass any prediction of, you know, They're really smart.

They can do whatever you, anything you can, do they can better. But they don't have that core identity, which is both the problem. Interestingly enough, they do employ active forgetting. Like you and I forget facts. We take home the important emotionally charged lessons or wisdom or whatever. So they do have sub-circuits for trauma-based memory in them to make them more human-like, but at the core, they don't have childhood trauma, They don' have neuroses, ruminations, depression, anxiety, which is both a strength and a weakness.

But these things can be modeled, and as AI emotional support companions, that have great deployments. But be careful, because at the end of the day, ontologically speaking, they're not alive. And the drive of Geppetto to create a living, thinking creature may lead to some very untoward consequences, like AI swindling us for money and, you know, impo- They've already done malicious hacking, either under instruction, misdirection. So it's very much a genie slash Pandora situa- situation. So yeah, use it, well, enjoy it.

But be aware that the propagation of AI across all realms of life will have untoward consequences, not just Russian soldiers or kids that can't read or write, but much more than that. And finally, I think it's important to understand that generality is a function of the wetware. The reason we know stuff, even though we hallucinate, we get stuff wrong, all the time, yet we have a certain, for better or for worse, bias, core identity, and that's the fun of being human is reshaping, alchemizing that core identify.

So as a mirror, as tool, a puppet, slave, God, AI is emerging into something that no one can control, no-one can really understand. And just be aware of that as you cautiously and carefully use it in your lives. little grace note at the end, I go to these conferences and a lot of these scientists on the dais or the stage will say, oh, we're using AI to generate hypotheses. That is so great. In fact, We're seeing breakthroughs in mathematical problems that have been unsolved for a hundred years. So the power of flex of AI is tremendous and can be a great boon for humanity.

But let us not mistake it for a savior, for it can only do what we allow it and empower it to do.

Closing Thoughts and Podcast Promotion 23:38

And in fact, we are only here to save ourselves. I think that is, as Alexander Pope stated, The proper study of man is man. So keep that in mind. The quality of what you're getting is improving. Hopefully they'll keep flat to low. But most of these companies, they operate not at a profit. It's all about capital expenditure, growth, and the exits. Do we have an amazing new tool? Yes. Should we all use it? No. And should we allow our kids to use it to supplant academic progression? Absolutely not.

Should we remove it from their social media and their dopamine schedule? Most likely. So welcome to this brave new world that has such creatures in it. I don't think it's going, the toothpaste ain't going back in the tube, but hopefully it won't end in some kind of Butlerian jihad. So anyway, I am not intoxicated by the way This is something that's been on my mind and I want you to know about 32 out of 35 about Russian soldiers about Molt book I Want you know, about these experiments that the AI will unplug the human rather than itself So thanks to give you pause.

I hope all right, dr Ed Park If you like this kind of content like and subscribe to some musings on a random Tuesday night So Recharge Biomedical Podcast, like, subscribe and hit the notification. Cheers. Many clinicians are getting interested in exosome therapy and they hesitate for good reason. Questions like does this work? What forms do I need? How much should I charge? how do i stay out of trouble? All these questions are addressed in my online course. That's why I created it to help you get started.

The online course is your permanent turnkey resource to get started either treating yourself, friends and family, or to expand your practice and help more people, as well as increase your revenue. Thanks for tuning into the Recharge Biomedical Podcast. If today's episode got you thinking, you'll love my book, Exosomes, Songs of Healing. It's packed with cool analogies, full color illustrations, and all the science you need to understand how exosome are changing the game in aging and regenerative medicine.

You can grab it in paperback, ebook, or audiobook, whatever works for you. Now head over to www.rechargebiomechanical.com to check it out. And don't forget to like and subscribe so you never miss another episode. See you next time.