37:04
creatine dosing
The 'Don't Die' team (Bryan Johnson, Kate Tolo, and Dr. Mike Min) argues that the large majority of disease — including most ER visits — is preventable through lifestyle and self-directed health measurement rather than reactive medicine, that patients should use LLMs like ChatGPT for medical self-advocacy, and that a widely-covered null study on creatine and muscle mass likely understated creatine's effects due to a too-short measurement window, with bodyweight-based dosing (~0.1g/kg/day) recommended over mega-dosing.
Emergency medicine is described as reactive — 'banging your head against the wall' fighting the secondary effects of a culture that doesn't value health, rather than addressing root causes.
90%+ of disease is claimed to be eventually preventable in general, and roughly 80% of ER cases specifically are described as preventable.
Acute conditions (trauma, appendicitis, infections) are hard to prevent, but chronic disease — the majority of what modern medicine treats — is largely preventable.
The US ER system is described as chaotic and overburdened compared to Australia and New Zealand; US doctors reportedly see ~30 patients/day versus ~10/day in New Zealand, a 3x effort differential.
Patients are encouraged to self-advocate by inputting their complaints into an LLM (ChatGPT) to get a probabilistic differential diagnosis and recommended exams, then push to have those exams done.
LLMs reportedly outperform physicians on standardized medical tests, but 'life is rarely standard,' limiting how far that generalizes.
Bryan describes using ChatGPT (a ~100-question intake) to correctly flag his mom's eye condition as inflammation/corneal scratch, later confirmed by an optometrist and resolved with steroid drops.
A recent study found 5g/day creatine for 12 weeks produced no muscle-mass advantage over training alone (both groups gained ~2kg lean muscle), and early creatine-related weight gain is attributed to water retention, not muscle.
The study's 7-day 'washin period' (used to control for fluid retention) is argued to be too short — actual muscle saturation at 5g/day reportedly takes 3–4 weeks, so the null result may be a methodology artifact rather than evidence creatine doesn't work.
Despite that study, creatine is described as having more supporting data than almost any other supplement, with positive findings for longevity, cognition, bone/muscle health, and metabolic health, and no studies showing harm.
Recommended dosing: ~0.1g/kg bodyweight/day (about 5–7g for an average adult) is enough to reach saturation naturally over 3–4 weeks without a loading phase; the classic loading protocol (20g/day for 5–7 days) is not necessary for general health goals.
Higher doses (10–20g/day) are used in cognitive-benefit studies, and up to 30g/day has been tested for up to 5 years without significant downsides — excess creatine is simply excreted in urine.
Vegans and low-meat-eaters are said to need more supplemental creatine since dietary intake (normally 1–2g/day from a standard diet) is lower for them.
Elevated creatinine on blood work from creatine supplementation is expected but not a reliable marker of dosing adequacy, since it's confounded by hydration and kidney function.
Years of personal health-protocol tracking is said to produce intense 'body awareness' (e.g., sensing one's own heart rate accurately), which can also create friction when encountering normal societal eating habits.
The body is described as highly adaptive, able to suppress major problems for a long time before a sudden cascade failure — illustrated via an inebriation study where the brain compensates and passes behavioral tests at low-to-medium intoxication despite scan-visible impairment, with compensation failing only at high intoxication.
Cognitive decline is claimed to begin long before symptoms appear and can, in principle, be detected early via brain scans before it manifests behaviorally.
Sleep score is said to correlate strongly with next-day work quality, creativity, and cognitive capacity, with effects accumulating across multiple nights rather than depending on any single night.
Kernel's stated mission is to measure the brain, reveal invisible neural activity, and pair that data with AI to accelerate improvements to human intelligence — including emotional development and blind-spot correction, not just IQ.
Bryan frames AI figures like Andrej Karpathy investing personal effort in sleep tracking and longevity as validation that optimizing one's own health/intellect is a sensible response to the current AI-driven moment.
LLM-based symptom intake / medical self-advocacy — Feeding personal symptoms into an LLM like ChatGPT to get a probabilistic differential diagnosis and a list of exams that should be run. Apply: Describe your complaints in detail to an LLM, ask what workup should be happening, and push your provider to actually perform the exams it suggests.
Weight-based creatine dosing (0.1 g/kg) — A bodyweight-proportional formula for setting daily creatine intake, e.g. ~7g/day for a 70kg person. Apply: Multiply your body weight in kilograms by 0.1 to get your target daily creatine dose in grams.
Gradual/no-loading creatine saturation protocol — Taking a steady ~5g/day of creatine without an initial loading phase, reaching full muscle saturation naturally after about 3–4 weeks. Apply: Start and maintain 5g/day consistently rather than front-loading; expect saturation to build over roughly a month.
Classic creatine loading phase — An older, higher-dose protocol of 20g/day for 5–7 days used to rapidly saturate muscle creatine stores. Apply: Reserve for cases needing fast saturation; not described as necessary for general health goals and comes with more water-retention effects.
Washin period (study-design technique) — A pre-study measurement window researchers use to account for fluid retention before assessing true muscle-mass change from creatine. Apply: When evaluating a creatine study's muscle-mass claims, check whether its washin period was long enough (the episode argues 3–4 weeks is needed at 5g/day) to isolate real gains from water weight.
Body-awareness via measurement-plus-subjective pairing — Combining objective biometric data with conscious attention to how you feel to build intuitive, well-calibrated knowledge of your own physiology. Apply: Pair wearable readouts (heart rate, sleep score) with deliberate awareness of your subjective state in the moment to sharpen your own health judgment over time.
Sleep-tracker comparative self-experiment — A personal two-month test comparing four sleep trackers (Aura, Whoop, Eight Sleep, Apple) against subjective work quality and cognitive performance, cited from Andrej Karpathy's own study. Apply: Run multiple sleep trackers side by side against a log of your daily output/mood to see which readings actually predict your real performance.
Biological age / aging-speed tracking — Measuring a pace-of-aging metric (e.g., reaching a 0.5x or slower aging speed) as a longevity outcome measure. Apply: Use available biological-age testing to check whether your interventions are slowing your measured aging pace over time.
Kernel brain measurement paired with AI — Kernel's approach of measuring brain activity to reveal invisible neural processes, then pairing that data with AI to accelerate improvements to intelligence, including emotional development, not just IQ. Apply: Use brain-measurement tools to surface personal blind spots, then combine the data with AI analysis to target specific areas for self-improvement.
Blueprint / Longevity Mix supplement protocol — Bryan Johnson's formulated longevity regimen, which bundles a 2.5g creatine dose inside its 'Longevity Mix' supplement. Apply: Cited as an example of a packaged multi-supplement daily stack rather than a standalone creatine-dosing method.
The claimed methodological flaw in the creatine study — a washin/measurement period roughly a quarter as long as the ~3–4 weeks actually needed for saturation at 5g/day — is presented as the likely reason for its high-profile null result, i.e., a headline-friendly negative finding attributed to study design rather than to creatine's efficacy.
Early weight gain from starting creatine is attributed specifically to water retention rather than added muscle, a distinction the speakers say is often conflated in casual interpretation of dosing results.
Dietary baseline matters for dosing: vegans/low-meat-eaters are said to need proportionally more supplemental creatine because they start from a lower baseline (~1-2g/day from a standard omnivorous diet).
Blood creatinine elevation, often read by patients as a kidney-function red flag, is described as an expected but unreliable proxy for creatine dosing adequacy because it's confounded by hydration status and baseline kidney function.
The inebriation-study analogy is used as evidence for a broader claim: the brain/body can mask real impairment (behaviorally normal, but scan-visible impairment) up to a threshold, then fail suddenly — offered as the rationale for scan-based early detection of cognitive decline before symptoms.
Practicing an intensive, measurement-driven health protocol is described as producing a kind of embodied precision (feeling one's exact heart rate) that simultaneously makes ordinary social/dietary norms feel more jarring by contrast.
Kernel's brain-measurement-plus-AI framing repositions personal health optimization as an existential-stakes project tied to navigating an AI-driven 'species transition,' not merely personal wellness — used to explain why an AI-industry figure investing in his own sleep/health is treated as validating evidence for the show's philosophy.
«I was just shocked at watching how the food arrived at various tables across the restaurant.»
— 00:09
«we believe we are the first generation who won't die and we are trying to practice that philosophy»
— 00:48
«Emergency medicine is just like banging your head against the wall all day long in the sense that you've got all of these issues in front of you that are secondary to a culture that doesn't value health.»
— 03:21
«I would throw your complaints into an LLM and ask it what should be happening and then make sure that that happens for yourself if I'm being totally honest.»
— 10:03
«Life is rarely standard.»
— 12:08
«Creatine has more data backing it than I think almost any other supplement that we're aware of. Um, and the data is overwhelmingly positive.»
— 17:50
«five grams a day seems adequate for the averagesized human. um given what the normal diet generally includes and that's going to get you to saturation. You don't need to do the the loading phase where you're taking 20 grams a day, you can just start at five and just do it consistently and then over 3 to 4 weeks you'll you'll get the load the load will happen naturally on its own.»
— 20:22
«as far as we can tell you take extra creatine you pee it out. It's not a big deal.»
— 21:25
«I know after doing my own health protocol for the past couple years, I'm intensely body aware. Like I I can feel my heart rate at any sec. I can tell you how fast or slow my heart is beating at any second. Um, with pretty good accuracy»
— 24:41
«The body is incredibly adaptive. That's what that tells me. It's like we can push off big problems for a very long time and then they can come crashing down.»
— 26:40
«cognitive decline begins a long time before symptoms emerge. And you can in fact do early detection of cognitive decline because your brain has impairment. You just don't see it like in your daily life.»
— 26:56
«my sleep scores correlate strongly with the quality of work I'm able to do that day. When my sleep score is low, I lack agency. I lack courage. I lack creativity. I'm simply tired. When my sleep score is high, I can power through anything.»
— 31:35
«we're evolving into a new species. And so my observation was that the most important thing we could do would be improving our own intelligence to see this moment for what it is and act wisely.»
— 33:49
«if we can measure the brain and reveal the invisible things that happen in the brain and then pair that with AI that will give us a vector to begin improving our intelligence»
— 34:08
Reception
Highly positive reception for the conversational format and hosts, with engaged audience providing constructive topic suggestions.
The episode is a loosely-structured but information-dense weekly debrief across three practitioners, translating a doctor's frustration with reactive ER medicine, a critique of a viral creatine study's methodology, and Kernel's brain-plus-AI thesis into concrete, appliable protocols (LLM-based self-advocacy, bodyweight-based creatine dosing) rather than abstract wellness advice.
37:04