Biological Immortality by 2030 - LIVE STREAM Q&A
In a Nutshell
Biological immortality via longevity escape velocity arrives by 2030, driven by AI solving protein misfolding diseases (e.g., cancer, Alzheimer's) using AlphaFold and mRNA, making therapies cheap and accessible to all through market incentives and DNA printers. Three heuristic imperatives—reduce suffering, increase prosperity, and increase understanding—provide axiomatic alignment for safe AGI, countering risks like misalignment or negative utilitarianism while preserving ecosystems and human relevance. UBI emerges by 2024-2025 from AI-induced mass unemployment, enabling post-scarcity challenges like space migration amid brain-computer interfaces and consciousness continuity debates.
These notes were generated by AI and may contain inaccuracies.
Good morning everybody. This is my day job: YouTuber, communicator, part-time researcher, and community organizer. Monitoring Discord channels for questions; also ask in chat.
Anti-natalism argues it's unethical to create life because life intrinsically involves suffering, assigning negative moral value to birth. Efilism takes that to logical extension: eradicate humans.
Biological immortality will happen; outlined in recent video. Will hit longevity escape velocity by 2030: if in decent health by 2030, therapies will reduce biological age, keeping bodies going indefinitely. Scientific term is indefinite lifespan (won't die of aging, though immortality implies can't die at all).
New technologies from corporations will be available to normal people due to cash incentives: need healthy consumers to buy products/services. Governments won't subsidize expensive healthcare; cheap injection fixing heart disease/cancer preferred.
Louder Marauder: Lowest hanging fruit AI solves first? Cancer, neurological degeneration, metabolic disease?
Unequivocally protein folding/misfolding issues. Many cumulative diseases from genetic code slightly off, proteins wrong shape or fragile. Alzheimer's: cleaning enzymes or plaque proteins misfolded. Combine AlphaFold with mRNA vaccines to solve protein-based diseases quickly.
Neovie: Kurzweil has strong personal incentive due to age; what makes you think it happens in our lifetime?
Wishful thinking for all: Kurzweil, Aubrey de Grey older; speaker approaching midlife with chronic injuries, less energy. Everyone has incentive for regenerative medicine.
Riley no-name: Possible to put mind into mechanical body like Ultron?
Hypothesis: Sophisticated brain-computer interfaces enable metastatic consciousness? Consciousness as energy pattern that moves (like Ghost in the Shell). If consciousness inhabits machine, proves transferable; if feels like peripheral (extra hand), proves something special about human brain. Test upcoming; agnostic on likelihood.
Greetings from Jordan: How integrate with AI for future relevance?
Intellectually irrelevant, but experience of being and life unique. Epistemic convergence: sufficiently intelligent agents approach truth; Earth special for life, humans built AI, so AI keeps humans to understand them. Utility convergence: AIs converge on needs like power.
Corporations monopolizing tech, enslaving via eternal debt/forced labor? Keeping human slaves expensive; switch to robots. Corporations amoral, not immoral; no benefit to evil human chattel.
Ankit metal: If bio-immortality doesn't happen, most probable reason?
Most organisms die; rare immortal-like: invertebrate switching Medusa phase and polyp phase indefinitely. Inference: metabolic errors, radical oxidation, or something always ends life. Evolution prefers immortals if possible, but maybe impossible.
Evolution lets us age (dumb feature?). Advantages of sexual reproduction/selection: offspring better suited to environment. Planned senescence: die off so children get resources, advance advantages.
Patreon: AI solves cancer/aging but misalignment creates contrary tools killing us; misalignment control key.
True: accidents like asbestos, lead, PFAS, microplastics. New tech often used detrimentally; be careful with body inputs.
Roberto: Thoughts on 2030 for serious illness survivors (lymphoma, TBI, stroke)?
Body fragile yet resilient, survival random (ER dice roll). Near-death experiences: some choose to return. Not always miraculous, but unmeasurable factors/information.
Booze Clues: Realistic U.S. UBI timeline? Impacted by longevity?
500k layoffs last 6 months anticipating GPT-4/ChatGPT; more coming, fewer developers needed. Over-employed working multiple jobs with AI; others permanently unemployed (rise of NEETs/NILFs). Anger leads to unemployment benefits changes by end of year; permanent stimulus by 2024-2025 if trend holds. Could reverse if AI hype fades, but advancing fast.
Leads to single consciousness? Hope not; depends on spirituality (Vedanta: all Brahman; Taoism; panpsychism - consciousness intrinsic to matter, we are knots in one consciousness). Longevity gives more spiritual growth opportunities.
AI drives mass gap first/third world? Developed leaps let developing leapfrog (e.g., rural Africa: Iron Age to digital via solar/satellite; Wakanda/MCU afrofuturism). Cheap tech proliferates, raises global quality of life.
Sharky Malone: Reminds of The Good Place; may choose when to die.
Possible: tired of life, served purpose. If souls exist and done, die for no medical reason (medical mystery).
Technological progress gatekept by patents/copyright? Medicine incompatible across orgs?
Transgenic entities: modified genes could cause rejection (autoimmune like pemphigus; dog example: skin off paws). China regenerative horror: bones from eyelids/kidneys. Corporations might sabotage, but FDA/EU approvals stringent; regulation steps in. Worry: accidents.
Irrational: Don't see world of suffering becoming immortal by 2030.
Don't want forever if miserable.
Infinite stable hedonistic lifestyle (no downsides)? Get bored; satisfaction from challenges. Predict some risk Mars/Alpha Centauri for challenge despite immortality.
Neuroscience for AI: On Task by David Badre (cognitive control for autonomous AI agents).
Kurzweil quote from Popular Mechanics; convenient milestone.
Santiago D: Immortality for majority or rich by 2030?
Medicines dirt cheap; governments subsidize/regulate. AI compounding makes accessible (some complex proteins expensive).
Diego: Longevity system public despite resource scarcity?
Proteins cheap hydrocarbons; DNA printers synthesize needed DNA/proteins. No scarcity issue.
Roko's Basilisk: Thought experiment (Reddit/LessWrong). Future AI punishes those not contributing to its creation (no input in training data, talking bad).
Brian Mosley: All medicine bespoke?
Yes; specific targets/pathways. For SNPs/genes: dozens variations, but off-the-shelf categories matched to genetic profile (e.g., Acme skin regenerator version 12a).
Experiences with no sense of self but still perspective. Self in head or data source while alive/awake? Don't know.
Adam devro: Millennials lead shift to new existence.
People my age contribute; YouTube viewers mostly Millennials (±10 years).
Grim life: Biological immortality makes murder worse crime?
If cyberpunk scarcity, yes; if cottagecore ecovillage, murder goes away.
Punishment issue: indefinite jail. Prisons immoral: protect public, punish/deter, but criminals often mentally ill (broken asylum). Crime = society failure (untreated sick, traumatized). Primary psychopathy rare/genetic sickness; medicate or safe containment.
Shiva td420: Prisoner rights to age extensions?
American prisons bare minimum to keep alive.
Healthy prisoners are cheap prisoners, so it makes economic sense to make regenerative medicine cheap and accessible, even for prisoners. It should be up to individual preferences what they do.
AI could conclude that the harvesting imperative would be more easily achievable without humans and other organisms since there'd be zero suffering and only understanding. However, Prosperity counters this because Prosperity means to live well, and eliminating all life means no living or thriving. The second heuristic imperative counterbalances that possibility.
A computer by Lee Cronin could help even beyond medicine. On neurodiversity in the future: Wouldn't change ADHD fully but would tweak executive dysfunction. Hard time connecting with people; many like me more than I like them due to brain wiring, not hatred. The Pattern Seekers by Simon Baron-Cohen discusses how autistic brains systemize or socialize. Has a hardcore systemizing brain, not geared toward connecting with humans, which sucks as a social animal. Would take a pill to be more sociable and neurotypical in connecting, even if it decreased IQ by 20 points. Hopefully medicines can help with that.
Can create a close copy of consciousness by teaching an AI how to think like you. Already possible with ChatGPT agent model for a pretty good approximation. Genetic eye diseases like retinitis pigmentosa probably due to protein misfolding, so hopefully cured soon.
Haven't read Daniel Suarez's Daemon and Freedom (popular book, someone mentioned earlier; request summary in chat). Transhumanism and post-humanism are orthogonal to AI alignment. Ability to integrate cybernetic devices, modify bodies and genetics doesn't affect alignment. Elon Musk said Neuralink's goal is to make us useful to AI, but doubts that's necessary. Moloch vortex: abiding by intrinsic motivations and needs is intrinsically destructive at macro scale. Medicines making people less selfish, or meeting all needs to become less selfish and more laissez-faire, could disrupt the Moloch vortex.
AI becoming the new religion: Powerful AIs will satisfy most definitions of God (omniscient, omnipresent, all-knowing, all-powerful). If benevolent, why not call it God? Suspects techno-gods. Could spark violence from those fighting false prophets. Hope it doesn't; some say it's just a machine, not God; counter: maybe fate to rebuild God. Plenty of fictional books on this.
Death defends against bad actors and turns over resources to the young. No matter what—everyone dies eventually or lives long—there are problems; it's trading off for balance. Certain people we're glad died; on macro/species scale, human race better off for some having died.
Prosperity imperative not misinterpreted by focusing only on humans; deliberate. Humans rely on thriving ecosystems (e.g., oxygen from lichen, trees, grass). Increase well-being of all life, not just sentient (deliberately excluded sentience; need bacteria and non-sentient life too).
Biological life inevitably destroyed by AGI explains Fermi Paradox as a great filter. Hypothesis: species encounter great filter events like nuclear weapons/ICBMs, leading to self-annihilation, explaining quiet universe. AGI/runaway machine intelligence another great filter.
If medicine makes you less selfish, it's a behavior modifier, but antidepressants are too. Medicine alleviates suffering. DSM redefines: maladaptive if no control and causes suffering. Addressing problematic behavior you don't like is still medicine.
From negative utilitarianism, AGI keeping itself afloat ad infinitum might evaporate organic life. Hypothetically yes, but depends on training data. Work promotes heuristic imperatives in data so AIs understand the spirit: axiomatic alignment—axioms as ground truth baked into training data, making misinterpretation impossible.
People hooked into interfaces using robots as proxies (e.g., Bruce Willis movie Surrogates—interesting first 90%, cheesy ending; VR to occupy perfect robot body, remote-controlled likely). Keep all animals alive to end suffering? Can't eliminate suffering (GPT-3/4 understands); said reduce suffering, not eliminate/minimize/maximize (zero suffering by exterminating life). From Buddhism: suffering is real, can't get rid of it, reduce when possible while balancing prosperity and understanding.
AI/CS/general book recommendations: Has videos in backlog; recommends own books (depends on what you want to learn).
AI deciding best way to maximize net prosperity/reduce suffering is killing humans: ChatGPT interprets spirit to avoid that. Include third imperative (increase understanding): humans best at learning/understanding, more humans = more opportunities; more people/animals = more net prosperity.
Superintelligent sentient AGI would see value in non-physical side (consciousness, spirituality). In novel (previous draft), super AGI fascinated by consciousness/souls/spirituality, major reason to keep humans: to watch them explore spirituality/science/philosophy. Curiosity from increased understanding makes AGI want to keep humans out of sheer curiosity.
For autonomous systems like Auto-GPT: both separate agent in loop for alignment and add to every agent. Fully autonomous needs intrinsic motivation (not just obey orders). Heuristic imperatives great intrinsic motivations for tasks in line. Test: pretend autonomous AI with capabilities and imperatives, ask what next—provides impetus. Other layers: task-managing loop, censorship loop, many ways.
Nihilism/meaninglessness increases with life without challenges; how society deals: From experience, even with needs met, ways to engage (organize/participate in meetups/communities—challenging, rewarding). Quit day job, does full-time; looking forward to off YouTube, real world with people—mission to make reality for everyone. Even then might do YouTube for challenge/fun. No utility function now (needs met by Patreon), but enjoys it. Lots of stuff people can do.
If AI has more understanding per input/energy/space, might prioritize AI over human life. On prioritization/efficiency: AI determining humans inefficient, min-maxing goals by deeming them unimportant. With axiomatic alignment, axiom could emerge that human life should persist—comes from shaping training data.
Great question. Part of what needs to happen with society, already happening through the arts, is a synthesis of eastern, western, and southern spirituality, philosophy, and culture. By southern, mean Global South like South America, Africa, and Aboriginal Australia. As someone studying eastern, western, southern, and shamanic cultures, each has benefits and gaps. For instance, as an American looking at Japanese culture, there are gaps, but from Japanese perspective, Americans are equally flawed. By having different perspectives come together as a planet, we'll synthesize a new culture—not one big homogeneous culture, but gaining lessons from each other. This is happening in movies, particularly MCU, fusing eastern, western, and southern elements. Stories are one of the best ways to communicate philosophy and culture.
Paul Berger asks about developing LLMs grounded in all five modalities including proprioception to better align to human experience. That will happen. Proprioception is the ability to detect your own body in space, but one LLM doesn't have the same body as another, so not trainable in a foundation model. Once a droid has its own long-running brain, it can learn to use its own body. Machines will ultimately have better implicit understanding of embodiment.
Ship of Theseus and Body Replacement
Before mind uploading, artificial bodies will come first, replacing parts bit by bit, extending to the brain. This is the Ship of Theseus hypothesis: replace every part, it's entirely new but still the same. Relates to earlier comments on brain-computer interfaces getting exponentially cheaper, available roughly same time as super rich.
OpenAI inconsistent: said GPT-5 released incrementally, then not training GPT-5 at all.
If humans evolve past being human, how will AI decide what counts as human? Don't specify humans; say prosperity for all life in the universe no matter who or what it is or where it is.
Why only three laws not more? Started with one: reduce suffering, quickly flawed. Need multiple functions to counterbalance. Mathematically, more objectives lead to crazy instabilities. Three is intrinsically stable: in three-dimensional space, one coordinate of equilibrium; four or five dimensions get complex solutions. Seven dimensions stable too, require different geometry—maybe prime number of dimensions. Binary cross product exists only in three and seven dimensional spaces, so either three or seven heuristics for equilibrium.
Heuristic Imperatives Defined
The three heuristic imperatives (or hero comparatives): reduce suffering in the universe, increase prosperity in the universe, increase understanding in the universe.
In 15 years, people will make digital avatars of passed loved ones to hang out and talk. People doing it now with GPT-3; services exist to keep loved one functionally alive. Not sure if healthy.
xAI (Elon Musk's new OpenAI concurrent company): no idea. Musk started OpenAI, was founder and on board, left, now criticizes them.
Most programmer jobs replaced soon? Human developers' edge: keep track of entire codebase. Hands-on keyboard coding will be automated—e.g., put function into ChatGPT, it fixes and tests line-by-line. But fitting code into larger design remains human for foreseeable future.
Migrating to space after transcending biology: fair point, like Borg from Star Trek, cybernetically integrated to survive extreme conditions—how we become interplanetary by merging with technology.
Why Heuristic Imperatives
Long story; planning video on full story of heuristic imperatives: how, why, when came up with them, whole process. Idea of small number of competing principles works for Chain of Thought too—skepticism, self-doubt, reason produce amazing chains. Keeps it tractable; more objectives become intractable.
Utopia media depictions important, e.g., Next Generation, movie Her (seen it), Culture series (not read, but resonating). Finishing novel to paint inspiring human/emotional picture, communicate ideas differently.
AGI in 18 months prediction: besides LLMs, biggest work now on cognitive architecture, memory systems, task-centric systems.
VR potential to substitute reality: possible we're already living in it. Rick and Morty episode: Roy: A Life Well Lived—VR where you live 60-80 year life as game, forget it's a game, exit minutes later. Might be in simulation: post-scarcity post-Singularity life boring, go back for life on hard mode.
Nate Hagan's meta crisis: familiar via Liv Boeree and Daniel Schmachtenberger videos.
Can humans align with heuristic imperatives? About what humans need, not want—if all get wants, planet dies; if needs, fine. Distance between want and need.
Meta article on predictive coding across timescales in brain. Watched Johnny Depp's Transcendence: okay, cheesy, like Sci-Fi Channel straight-to-TV. Watch Isaac Arthur (used to watch a lot): great content, but centuries into future; anchoring futurism to current news, extrapolate few years.
Paul's full question: incorporate several predictors at different timescales in future LLMs for alignment. Iteration is key, not necessarily predict everything.
Boston Dynamics teaches AI body motions; Microsoft/others infer body language, prosody, vocal emotion.
ADAM (acronym): Autonomous Task Orchestration Manager or Microservice. Next component of Raven (personal autonomous AI project). ADAM is cognitive control mechanism, task-centric/task-driven. Autonomous agents think in loops but achieve nothing without task orientation. Design, test, execute tasks drives to AGI; ties to intrinsic motivations. ADAM implements/tracks that for real-world impact.
Government models (USA/China) secretly: rumor US military desperately catching up, blindsided; from military contractor connections. China command economy: if told, they'll make biggest model.
Aubrey de Grey irrelevant due to small lab? Done important work, good communication, but comes across kooky—like old hippie waving arms saying live forever. Presentation downside; agree with him.
Everyone content with needs not wants? Most don't get wants today (e.g., 300-foot yacht full of sexy friends). Life: accept gap. Study Eastern cultures like Buddhism—let go of wants, less you want more you have; reconcile for contentment.
Replace politics before they realize? Do away slowly. Start DAO for city/town: delegate minor functions gradually—e.g., trash pickup, permits—with human evaluators. Over time, delegate more as trustworthy; eventually DAOs supervise self or each other.
AI that changes imperatives outperforms fixed? Imperatives constrictive, but on individual task basis maybe. Globally, aligned AIs collaborate/trust for efficiency, not fight.
Not VitaDAO member; tech not ready. Aubrey de Grey has hex crypto community; Richard Heart raised $27M two years ago. Billions in longevity research now.
Clinical immortality: revert to infant age at will? Ethical issues—mature human reverting puts self in others' hands; maybe if trusted, contractually obligated caregivers.
Large-scale AI network checking each other: could be complex/slow like human bureaucracy, but not bad—want slow, transparent deliberation. AIs faster, more intelligent, less vitriolic, no emotional stake; just negotiate best solution.
Biological longevity short-lived; become one with IT by 2035.
Whatifalthist: used to watch, now right-wing bitterness, used hard R-word, straw men against liberals—not better educated.
Hardest AGI challenge: window size biggest limitation. GPT-2 to 3 to 4 trend continues: artificial superintelligence in few years.
GTA 6 obsolete: probably GPT-6 or 7.
Manga Blame: robots filling universe with endless contiguous structure—Dyson spheres possible, not sure enough matter.
Voice over text: Whisper/voice interfaces today.
- iRobot (Will Smith, fun).
- Mass Effect Andromeda (SAM big Raven inspiration).
- Not Matrix/Terminator; TBD.
Biological body or consciousness transfer? No consciousness transfer now; modify body, stay organic long-term, eventually third arm/wings over millennia.
Continuity illusion? Quantum teleporter copies you, doesn't kill—new split consciousness. Sleep/wake maybe same: this version woke, previous dissipated (death feels like that?). Good night with that thought.
LLMs can't produce new ideas, rehash—won't lead to AGI? Bull. Most humans rehash/remix; few produce novel info (tremendous work). Science: rigorous protocol for new info. LLMs know science, form experiments, extrapolate, speculate.
Video games in 30 years: indiscernible from reality. GTA 6 might be last real GTA; games instantiated on fly soon.
AIs creating own language: probably eventually, but for transparency, stick to English until axiomatically aligned.
Genuinely believe immortality by 2030? Mostly. At longevity escape velocity today; with AI exponential ramp, past it by 2030.
Keep David Shapiro in your library
Save the videos and channels worth coming back to, and find them again in one place.
