Transcript 0:00 [gentle music] "This is the way the world ends, not with a bang, but a whimper." T.S. Eliot wrote that line a century ago. 0:14 But philosopher Nir Eyskovitz thinks it might be the perfect description of how artificial intelligence ends humanity. 0:23 In 2023, Eyskovitz published an article titled AI Is an Existential Threat, Just Not the Way You Think. His argument wasn't about killer robots or civilizational collapse. It was about something stranger and quieter, 0:42 the possibility that we simply gradually stop being human, one machine upgrade at a time, until the day arrives when we have replaced ourselves entirely. There's no sudden events, no dramatic last stand. 0:59 We just disappear. Let me show you how it could begin. [gentle music] I wake up at 6:47 AM. I know this because I know it, not because I checked a clock. 1:18 The time appears in my awareness the same way the color of the ceiling does. My neural interface registered the moment my consciousness surfaced from sleep and logged it automatically. 1:29 I think about coffee, and by the time I reach the kitchen, it's brewing. I don't need buttons, voice activation, or gesture control, just my thoughts seamlessly integrated with the systems around me. 1:45 This is me in 2040, though I'm not sure what me means anymore. About 12% of my cognitive processing now happens outside my biological brain. 1:58 My hippocampus is supplemented by a neural prosthetic that handles most of my long-term memory encoding. My prefrontal cortex is augmented with a processing layer that helps me focus, plan, and resist impulses. 2:13 When I do mental arithmetic, I genuinely don't know whether the calculation is happening in my neurons or my implants. I'm not even considered heavily augmented by contemporary standards. 2:28 About two billion people worldwide have some form of cognitive enhancement, and the number grows by about 100 million every year. My son thinks my generation is quaint for still having mostly biological brains. 2:43 He got his first implant at 16. By the time he's my age, he might not have any brain tissue at all. He'll still be my son, won't he? 2:57 In part one of this series, I covered how super intelligence could decisively wipe us out. In part two, how accumulated AI risks could lead to civilizational collapse. 3:12 This is part three, the possibility that neither happens, and we lose humanity by default. It's more philosophical and existential, and perhaps my waking up in 2040 wasn't the beginning either. 3:29 The takeover has already begun. [gentle music] This is Eyskovitz's argument. AI won't kill us. It won't collapse our civilization. 3:46 It will erode the abilities and experiences that make us human in the first place, human traits like judgment, serendipity, and critical thinking, the capacity to do things for ourselves. 3:59 The philosopher Daniel Dennett came to a similar conclusion. "AI tools that we become hyper-dependent on, that is going to happen, and one of the dangers is that we will give them more authority than they warrant." 4:13 Consider what we've already outsourced: arithmetic to calculators, spelling to autocorrect, navigation to GPS, memory to our phones. Each step was more convenient, so we didn't resist. 4:28 This is something I've explored in an earlier video about how we've already become cyborgs without really noticing. "We're judgment-making creatures," Eyskovitz argues. 4:40 We rationally weigh options and make daily calls about whom to trust, what to prioritize, and how to act. But more and more of these judgments are being automated and farmed out to algorithms. 4:53 The fewer we make, the worse we become at making them. We value serendipitous encounters, coming across a book, a person, or an idea by accident and being drawn into it. 5:07 But algorithmic recommendation engines exist to eliminate that serendipity, replacing chance with prediction. The unexpected encounter becomes rarer as systems get better at giving us exactly what we already want, 5:22 and our critical thinking is atrophying. 5:25 In educational settings, the ability to reason through problems, hold complex arguments in working memory, and evaluate evidence, these skills are becoming optional when you can simply ask an AI. 5:38 In June 2025, researchers at MIT published a study called Your Brain on ChatGPT. 5:45 They measured the brain activity of students writing essays, some using AI, some using search engines, and some using nothing but their own minds. 5:56 Students who used ChatGPT showed significantly weaker neural connectivity across all frequency bands.Their brains were doing less work, and when asked to quote from essays they had written just minutes earlier, eighty-three percent of AI users couldn't do it accurately. 6:12 The brain-only group failed only eleven percent of the time. The researchers called it cognitive debt, the accumulated cost of outsourcing your thinking. 6:22 When the AI users were later asked to write without assistance, their neural patterns didn't recover. 6:28 They showed what the researchers termed cognitive deficiency, weaker brain engagement than people who had never used chatbots at all. And the MIT findings aren't isolated. 6:39 A 2024 study on AI and learning found that while ChatGPT improves short-term task performance, it produced no significant gains in actual knowledge acquisition or transfer. 6:51 The researchers called it metacognitive laziness. Students offload their thinking to AI and stop regulating their own learning. They feel more confident, but they haven't actually learned anything. 7:04 Another study found cognitive offloading to AI strongly correlates with reduced critical thinking, with a correlation coefficient of minus naught point seven five. 7:14 The more you trust the tool, the less you engage your own analytical faculties. The MIT study opens with a quote from Frank Herbert's Dune, written in 1965. 7:26 "Once men turned their thinking over to machines in the hope that this would set them free, but that only permitted other men with machines to enslave them." We're not being killed, we're being hollowed out. 7:41 But cognitive erosion is only the first phase. The replacement is likely to become more physical. 7:48 In his 2023 book, The Coming Wave, Mustafa Suleyman argues that we're approaching a convergence of artificial intelligence and synthetic biology that could reshape what it means to be human. 8:00 Ray Kurzweil has been making similar predictions for decades. He believes we'll have nanobots in our brains connecting us to the cloud at some point in the 2030s. 8:10 This oncoming technological threshold is often referred to as the singularity, the point at which artificial intelligence becomes capable of recursive self-improvement, getting smarter, then using that intelligence to get smarter still in a loop that accelerates beyond our ability to predict or control. 8:30 If that happens, the technologies Suleyman describes won't arrive gradually over decades. They'll arrive very suddenly, without much time to adjust or regulate. 8:41 We already have cochlear implants sending digital audio directly into the brains of over four hundred thousand people worldwide. 8:49 Neuralink has implanted chips that let paralyzed patients control computers with their thoughts. Bionic limbs connect to nerves and restore the sense of touch. 9:00 Exoskeletons give workers and rehabilitation patients strength they wouldn't otherwise have. These are medical interventions developed to restore lost function. But the trajectory is clear. 9:14 The same technologies that restore can enhance. And Suleyman's central warning is that these technologies cannot be contained. Once they exist, they proliferate. 9:25 Once enhancement becomes possible, competitive pressure makes it difficult to refuse. The question shifts from are we outsourcing our thinking to something more fundamental. 9:37 If we start replacing biological components with synthetic ones, at what point do we stop being ourselves? The ancient Greek historian Plutarch described a paradox that philosophers have debated ever since. 9:51 The ship of the legendary hero Theseus was preserved in Athens as a memorial. Over the years, as planks rotted and decayed, the Athenians replaced them with new timber. A plank here, a mast there. 10:04 Eventually, every original plank had been replaced. The question, was it still the ship of Theseus? Your intuition probably says yes. The ship maintained its form, its function, and its continuity. 10:20 At no single moment did it stop being the ship and become something else. The replacements were gradual, each one preserving the overall structure. 10:29 It carried the same name and held the same meaning for the Athenians who preserved it. But the philosopher Thomas Hobbes added a complication. 10:38 What if someone had collected all the discarded original planks and built a second ship from them? Now you have two vessels. One has been continuously maintained but contains no original material. 10:50 The other is built entirely from original parts but has no continuity of existence. Which one is the real ship of Theseus? There's no satisfying answer. 11:02 The paradox exposes something uncomfortable about how we think about identity. We assume things persist over time as themselves. 11:10 But if every component can be replaced while identity supposedly remains, what exactly is identity made of? This paradox becomes urgent when applied to the brain. Imagine a near-future medical procedure. 11:25 A single neuron in your prefrontal cortex is damaged. Doctors replace it with a synthetic equivalent, identical in function, connecting to the same neighbors, processing signals in exactly the same way. 11:38 You notice nothing. You feel exactly the same. One neuron seems harmless, so when another fails, you replace that too. And another. Over years, as biological neurons wear out, synthetic ones take their place. 11:55 Each individual replacement preserves your sense of continuous identity. At no single moment do you feel like you've become someone else.But eventually, every neuron has been swapped. Your brain is entirely synthetic. 12:09 The question is no longer philosophical. Are you still you? Of course you're still you. Your memories are intact. Your personality is unchanged. You have continuous experience throughout the process. 12:25 In his twenty twenty-four book, The Singularity Is Nearer, Ray Kurzweil offers a direct answer to the ship of Theseus problem, and it begins with an observation about your brain right now. 12:36 You think of your neurons as stable things, the physical foundation of your identity. But the brain is in constant flux at the molecular level. 12:45 The mitochondria in your neurons turn over about half their components every month. The proteins that power your synapses are replenished every two to five days. 12:55 The NMDA receptors, crucial for learning and memory, are replaced in a matter of hours. Our brains are almost completely replaced within a few months. 13:05 You are already biologically a different physical entity than you were last year, yet you feel like the same person. Your memories persist. Your personality continues. This is Kurzweil's key point. 13:20 If natural biological replacement preserves identity, why should synthetic replacement be any different? We're not our atoms. 13:28 We're patterns, information structures that persist even as the physical substrate flows through us like water through a river. And Kurzweil distinguishes between two scenarios. 13:40 In the first, someone makes a complete copy of your brain. Is the copy you? Kurzweil says no. The moment of copying creates divergence. 13:51 But in the second scenario, we gradually replace each section of your brain with a digital equivalent connected to your remaining neurons throughout. There's no moment of duplication, no divergence. 14:04 After every replacement, you still feel like you. The difference is continuity. 14:10 The ship with institutional continuity is the real ship because identity tracks a continuous process of maintenance, not the material substrate. Perhaps consciousness is substrate independent. 14:23 Perhaps the enhanced being genuinely experiences life as richly as the original. The lights could stay on throughout the process. But this confidence rests on an assumption that might be wrong. 14:37 The philosopher David Chalmers, famous for coining the term the hard problem of consciousness, has spent decades considering precisely this question. 14:45 And to understand his answer, we need to understand a distinction he draws between two different meanings of the word consciousness. The first is functional. 14:55 It refers to the ability to be aware of your surroundings, to distinguish your internal thoughts from the external world, and to respond appropriately to stimuli. 15:05 By this definition, a deeply sleeping person is not conscious. A drunk person is partially conscious, and a sober alert person is fully conscious. This kind of consciousness can generally be observed from the outside. 15:19 We can measure it. Brain scans can detect it. 15:22 But there's a second meaning that's more relevant to questions of identity, the ability to have subjective experiences inside a mind and not merely to give the outward appearance of doing so. 15:33 Philosophers call these subjective experiences qualia, 15:37 the redness of red, the painfulness of pain, the felt quality of experience that makes consciousness something it is like to be rather than just a functional process. But we cannot detect qualia from the outside. 15:53 Even assuming that other humans have subjective experiences is technically just an assumption. 15:59 We make it because other people's behavior resembles our own, and we know we have inner experiences, but we can't verify it directly. 16:08 Chalmers illustrates this through a thought experiment about what he calls philosophical zombies. 16:14 Not the shambling undead, but hypothetical beings who show all the detectable neurological and behavioral signs of consciousness while having no subjective experience whatsoever. 16:26 A zombie would pass every test we could devise. It would claim to see red, describe the experience of pain, and write poetry about the feeling of love. But inside, nothing. There's no qualia or inner light. 16:40 Science could never tell the difference between a zombie and a normal human. 16:45 If such zombies are even theoretically possible, Chalmers argues, then there's no necessary causal connection between the physical systems that process information and the subjective experiences that seem to accompany them. 16:59 Consciousness might be something separate from function, something that could, in principle, be present or absent regardless of behavior. 17:08 The problem is that phenomenal consciousness can only be experienced from the inside. There is no external test for it. We cannot prove silicon can be conscious. We cannot prove it can't. 17:19 And the being on the other side of the transition would have no way to tell us whether the lights were still on because telling us is just a function, not experience. 17:30 A simulation of consciousness, no matter how sophisticated, might not actually be conscious. There's a related insight from Hans Moravec, the roboticist and AI pioneer. 17:42 Moravec is actually an enthusiast for mind uploading. But in thinking about how to build intelligent machines, he identified something strange about biological intelligence, something now called Moravec's paradox. 17:55 The paradox is this. 17:57 Mental tasks that seem hard to humans, like calculating square roots or memorizing long strings of numbers, are comparatively easy for computers.But mental tasks that feel effortless to us, like recognizing a face, walking without falling over, or understanding a joke, are enormously difficult to program. 18:16 The reason, Moravec argued, is evolutionary depth. The skills we find easy, like perception and movement, have been refined over hundreds of millions of years of evolution. 18:27 They run in the background of our brains, automatic and efficient. 18:32 The skills we find hard, like abstract reasoning and mathematics, are powered by the neocortex, which only reached its modern form several hundred thousand years ago. 18:41 The ancient skills are more deeply embedded in our biology. The recent skills run on newer, more flexible, but also more fragile cognitive machinery. This has implications for gradual enhancement. 18:54 The parts of cognition that are easiest to augment or replace, memory, calculation, focus, are the recent high-level functions. 19:03 The parts that are hardest to replicate are the ancient, embodied skills that depend on millions of years of fine-tuning. But those ancient systems aren't just about movement and perception. 19:15 They're also involved in emotion, intuition, and the gut feelings that guide our decisions in ways we don't consciously understand. Replace the neocortex with silicon and you might preserve abstract thought. 19:28 But what about the deeper systems? What about the felt sense of being a body in the world? The emotional coloring that makes experience meaningful? The neuroscientist Miguel Nicolelis puts it directly. 19:43 The brain is not computable. Consciousness emerges from non-linear, unpredictable interactions of billions of neurons. 19:51 Embedded in a body, bathed in hormones, connected to gut bacteria, and responsive to the immune system, it's not a program that can be ported to new hardware. 20:01 It's a process that depends on specific physical properties of biological neurons. John Searle's Chinese room argument makes a similar point. 20:11 A person who doesn't speak Chinese could sit in a room, receive Chinese characters, look up the appropriate response in a rulebook, and send back Chinese characters that form a coherent reply. 20:23 To an outside observer, the room appears to understand Chinese. But the person inside understands nothing. They're just manipulating symbols according to rules. 20:34 Silicon, Searle argues, can only ever do symbolic manipulation. It can process syntax but not semantics. It can mimic understanding without possessing it. 20:45 If Nicolelis and Searle are right, then the gradually upgraded brain doesn't preserve identity at all. 20:52 It creates an elaborate puppet, a system that acts as if it's conscious, claims to be conscious, and has no idea that the lights have gone out. 21:01 The original person died somewhere during the process, perhaps when the hippocampus was replaced, perhaps when the emotional centers were swapped for silicon. We would never know. 21:13 Neither would the being that replaced us. In 2017, the historian Yuval Noah Harari published Homo Deus, a brief history of tomorrow. 21:24 His thesis is that humanity, having largely conquered the traditional scourges of famine, plague, and war, is now turning to a new agenda. Three divine goals. 21:34 The first is immortality, the conquest of death through biotechnology. If aging is a disease, it can be cured. If death is a technical problem, it can be solved. 21:47 Google's Calico Labs and dozens of longevity startups are pursuing this with billions in funding. The second is bliss, the engineering of perpetual happiness. We already manipulate mood with pharmaceuticals. 22:02 Future technologies could optimize brain chemistry continuously, eliminating suffering and maximizing pleasure. Why feel sad when sadness is just a chemical state that can be adjusted? 22:15 The third is divinity, the upgrading of homo sapiens into something greater. Enhanced intelligence. Expanded consciousness. Powers that would seem godlike to our ancestors. Those goals seem rational, perhaps even noble. 22:32 Who wouldn't want to live longer, suffer less, and reach higher? But Harari identifies a self-destructive paradox at the heart of this project. 22:42 The same technologies that can upgrade humans into gods might also make humans irrelevant. The enhancement won't be universal. Harari is explicit about this. 22:53 Billions of people in developing countries and neglected communities will continue dealing with poverty, illness, and violence, even as the elites are already reaching for eternal youth and godlike powers. 23:06 A massive new class emerges. Harari calls them economically useless people, made redundant by AI automation and unable to access enhancement technologies that might make them competitive again. 23:21 Meanwhile, the enhanced elite face their own trap. Competitive pressure makes enhancement mandatory. To stay relevant in business, academia, politics, and even social life, you need the upgrades. 23:35 Each enhancement is individually rational. Collectively, they push humanity further from its biological baseline. The pursuit of immortality leads to technologies that make biological bodies optional. 23:47 The pursuit of bliss leads to simulated pleasures that make physical reality feel dull and limited. The pursuit of divinity leads to cognitive powers that make ordinary human concerns seem trivial. 24:01 As Harari puts it, while the attempt to upgrade humans into gods takes humanism to its logical conclusion, it simultaneously exposes humanism's inherent flaws.We become gods. We also become obsolete. 24:16 The category human empties of meaning just as we claim to be perfecting it. There's one more exit from humanity. We might simply leave physical reality behind. 24:28 As virtual environments improve, and they will improve dramatically, especially post-singularity, digital existence starts to look appealing. 24:38 Your mind preserved on servers that can be maintained indefinitely, backed up, free from aging and disease, potentially immortal in a way biological bodies can never be. 24:51 The environment can be optimized for pleasure, beauty, or stimulation, precisely calibrated to your preferences. You could fly, visit ancient Rome, explore the surface of Mars. 25:04 Why would anyone stay in a decaying biological body when the alternative offers so much more? This is a premise at the core of the movie franchise Avatar, the highest grossing of all time. 25:16 Research on the digital afterlife describes AI cloud consciousness, the uploading of minds into digital realms where they can exist indefinitely. 25:25 Companies like Hereafter AI and Eternym are already creating digital avatars of deceased individuals. The technology is crude now. It won't stay that way. 25:37 If the virtual world becomes better than reality, more pleasurable, more interesting, more controllable, rational beings will migrate to it. Physical bodies become inconveniences, then burdens to be shed. 25:52 The physical world empties out. The uploaded consciousness will insist it's still the same person. It has the memories, the personality, and the sense of continuous identity. 26:02 But it has no experience of rain on skin, the taste of food, or physical touch. It has never inhabited a body that evolved over millions of years in the physical world. It's a being that remembers being human. 26:16 That's a different thing from being human. So how does it end? The decisive risk scenario gives us superintelligence deciding we're inefficient and eliminating us with ruthless machine logic. 26:29 It's the science fiction scenario that seems more plausible by the day. But more probable is an accumulative risk scenario that leads to major dysfunction. AI amplifies our worst tendencies until society stops working. 26:44 We've seen through history how civilizations fall into dark ages. The existential risk scenario gives us something more philosophical. But perhaps if the other scenarios don't happen, it occurs by default. 26:57 Multiple pathways all leading to humanity being extinguished by something we created. If each generation becomes more dependent on AI and less capable of independent thought and judgment, then skills will atrophy. 27:12 And when we're weak enough, enhancement starts to look like a solution rather than a threat. So we upgrade. Biological neurons replaced by synthetic ones. 27:23 The ship of Theseus completes its voyage with no original planks remaining. Maybe consciousness survives the transition. Maybe it doesn't. We'll never know because the being on the other side will insist it's fine. 27:37 Or we retreat into simulated worlds. Why would we live in a physical universe with its harsh realities when we could escape into anything we want? Reality empties out. 27:49 And underneath all of this, global birth rates are already falling below replacement level across the developed world. South Korea's fertility rate has dropped below 0.8. 28:01 If the enhanced pursue immortality, why have children? If the uploaded exist in digital realms, why create physical offspring? Humanity doesn't need a catastrophe to end. It just needs to stop replenishing itself. 28:16 Leon Kass, who chaired the US President's Council on Bioethics under George W. Bush, wrote, In our attempt to become more than human, we will end up less than human. Or indeed, with no humans at all. 28:35 This is the way the world ends. Between the desire and the spasm. Between the potency and the existence. Between the essence and the descent. Falls the shadow. For thine is the kingdom. For thine is. Life is for. 28:53 Thine is the. This is the way the world ends. This is the way the world ends. This is the way the world ends. Not with a bang, but a whimper. 29:10 You've been watching Absolutely Agentic. We're a media consultancy and startup focused on AI, and we create weekly YouTube videos and a newsletter we send a few times a week. 29:20 In a world of AI hype, we're trying to strike a balance, looking at a range of opinions to cut through an increasingly complex world. 29:28 The best thing you can do to support us is to sign up to the newsletter in the description or subscribe to the channel. But also don't forget to like and leave us a friendly comment. See you next time. 29:40 It'll be Absolutely Agentic.
