Johnson, Bryan

founder of Kernel, a non-invasive brain-interface company; previously founder of Braintree/Venmo

Bryan Johnson: Kernel Brain-Computer Interfaces

Lex Fridman Podcast · May 24, 2021

• Bryan Johnson
• Kernel
• brain-computer interfaces
• non-invasive neuroimaging
• Neuralink comparison
• quantified self
• biomarker tracking
• zeroth principle thinking
• goal alignment
• consciousness engineering
• Braintree and Venmo
• mortality and depression

Core ideas

The episode opens with Lex wearing Kernel's "Flow" headset — a non-invasive functional near-infrared spectroscopy device using 52 laser-and-sensor modules to measure blood-oxygenation changes across roughly a thousand channels as a proxy for neural activity — while Johnson tells jokes to demonstrate real-time cortical measurement. Johnson frames brain interfaces primarily as measurement systems rather than control systems (the popular "move a cursor with your mind" framing, in his view, undersells the technology), arguing that once brain activity can be quantified at scale, an open-ended range of applications becomes possible, much as wearables turned heart rate and sleep into actionable personal data. He argues human cognition — internal thought, conversation, attention — is arguably the largest raw "input material" into society, yet remains almost entirely unquantified and lacks any formal engineering discipline, unlike nearly everything else modern society measures (distances, calories, genomes); he sees introducing that formal measurement layer as Kernel's most significant potential contribution.

Johnson describes Kernel's business strategy as explicitly avoiding a traditional "killer app" search in favor of what he analogizes to the Drake equation: building technology that meets thresholds of data quality, cost, and comfort, then letting mass adoption and organic experimentation (by both experts and non-experts) generate discoveries, similar to the early internet's ecosystem-driven growth rather than a single predetermined use case. He recounts funding Kernel's first $53 million himself after struggling to secure investors (228 pitched, one said yes) because there was no known "killer app" for a fundamentally new measurement category. He discusses an internal Kernel study in which his own device data showed his prior night's sleep quality independently predicted his impulse control the next day — offered as an example of the kind of personally actionable, otherwise-invisible correlation continuous brain measurement could surface, with potential applications including work/focus optimization, meditation, psychedelic research (arguing psychedelics research currently lacks the kind of quantified baseline/before/after measurement standard in other areas of medicine), and mental health more broadly (contrasting the vague, self-reported nature of current depression/suicidality assessment with the precision of, e.g., cardiology).

On Kernel's technical approach versus competitors, Johnson recounts early conversations with Elon Musk (who also considered joining Kernel before founding Neuralink) about pursuing brain interfaces, noting Kernel initially pursued invasive technology before pivoting to non-invasive approaches after conversations with veteran neurosurgeons and companies with decades of experience bringing implants to market; he frames Kernel's non-invasive "Flow" and "Flux" devices and Neuralink's invasive approach as parallel, genuinely uncertain bets on which technological path will eventually reach mainstream adoption, comparing the current landscape to dial-up-era internet bandwidth constraints, and estimates both companies have separately spent over $100 million just to make a credible attempt.

Johnson explains his personal biography — leaving a repressive religious upbringing, a decade of chronic depression, and later founding the payments company Braintree specifically to accumulate enough capital ("enough money by 30 to never have to work again") to fund a subsequent, purely mission-driven project — describing in detail his door-to-door credit card processing sales tactics as a young entrepreneur and his strategic acquisition of Venmo to replicate PayPal's two-sided-marketplace advantage before Braintree's sale to PayPal/eBay. He connects this arc to his current approach to health and decision-making, describing an extensive personal regimen (testing over 200 biomarkers every 90 days, eating a fixed vegan diet timed to protect sleep quality, avoiding self-report of momentary feelings in favor of longitudinal biological data) which he frames as deliberately "demoting" his conscious mind from decision-making authority — arguing his conscious mind has repeatedly been shown (through his own experience with religion, depression, and cognitive biases) to be an unreliable narrator of what is actually good for him, and citing dramatic personal improvements in health markers since ceding dietary control to data.

He extends this "goal alignment" framing to a broader thesis: that the cost of designing, manufacturing, and distributing intelligence is trending toward zero (as evolution's biological design process is supplemented by human-engineered intelligence), and that this makes the central question of existence increasingly one of negotiating "goal alignment" among the many biological, psychological, and eventually artificial agents that constitute a person and a society, rather than a single unified conscious will. On AI risk, he acknowledges the instinctive fear response to novel, powerful technology but argues this generation is uniquely positioned to aspire toward entirely new, currently unimaginable forms of consciousness or existence, which he considers underexplored relative to purely defensive, risk-avoidance framing. He also discusses a personal framework he calls "zeroth principle thinking" (distinct from first-principles thinking) — the deliberate practice of taking seriously ideas so revolutionary that their consequences can't yet be reasoned about, illustrated by AlphaGo's moves appearing "alien" to expert Go players and by his own realization (via the book Zero: The Biography of a Dangerous Idea) that the number zero itself was once a civilization-altering, initially unimaginable concept — arguing that as AI increasingly participates in idea generation, "zero-level" (unprecedented) breakthroughs will likely accelerate.

On consciousness, Johnson is sympathetic to an "engineering the illusion of consciousness" framing (agreeing with Lex that a functional, socially convincing display of consciousness may be more tractable and useful to pursue than resolving the philosophical "hard problem"), and speculates that sufficiently rich brain-interface data could let two people share something closer to a quantified, shared vector-space representation of an emotional state (like happiness) rather than relying on words and inference via theory of mind — while also raising this as a route to more deliberate, quantifiable, and consensual "consciousness engineering" between humans, and eventually between humans and AI systems trained with human-behavioral data to interact more appropriately. On data privacy, he argues the internet-era norm (companies extracting maximal personal data without transparent consent or control) was a bad starting precedent, and that brain-interface data — being especially personal — requires establishing individual consent, transparency, and deletion rights as foundational, citing Kernel's app design (individual control over one's own recorded data) as a deliberate attempt to set a better precedent.

The episode closes on a personal register: Johnson recounts a harrowing 2016 Kilimanjaro summit attempt during a period of severe depression, divorce, and religious departure — including altitude sickness, stomach flu, and a guide's simple instruction ("look up, take a step, take a breath") that he credits as a turning point toward rebuilding his life — and offers advice on evaluating others' advice by examining the "assumption stack" underlying it rather than the advice itself, since most advice reflects a rapidly depreciating prior context rather than durable truth. Asked about the meaning of life, he frames it (echoing his broader thesis) as goal alignment amid accelerating intelligence, and cites James Carse's "finite and infinite games" as his closing framework: the aim, in his view, is to remain positioned to keep playing an open-ended game rather than to win any particular one.

Excerpts