In this interview, Stanford neuroscientist Andrew Huberman argues that leading AI companies like OpenAI and Neuralink are essentially becoming biotech firms, aiming to non-invasively read and write brain states. He is optimistic about neurotechnology, especially an upcoming MIT sleep-inducing eye mask (launching in 7-12 months) and Neuralink's implant (its surgeon has a subcutaneous receiver in his hand to open doors). He also warns about gene-editing companies like Herocyte and Orchid, which are trying to link genes to IQ, posing risks.
Andrew Huberman (Stanford neuroscientist) discussed the future of neuroscience on the show, arguing that the field is shifting from "reading the brain" to "writing the brain"—achieving precise control over specific brain states. He noted that leading AI companies (such as OpenAI and Anthropic) are e
Andrew Huberman (Stanford neuroscientist, podcast host) discusses the future direction of neuroscience with Patrick O'Shaughnessy. Huberman argues that leading AI companies (OpenAI, Anthropic, Meta, Neuralink) are essentially evolving toward becoming biotech companies, with the ultimate goal of achieving non-invasive reading, writing, and control of the awake brain state. He believes that understanding and precisely regulating the awake brain state is the most pressing issue in neuroscience today, and that technologists are flooding into this field at scale, pushing brain-computer interfaces (BCI) beyond the "cyborg" boundary.
Huberman argues that for top talent, energy supply has replaced external competition as the core bottleneck, but blindly imitating extreme cases (e.g., Elon Musk's low sleep needs) is dangerous.
1. Honestly assess sleep needs: Most people require 6–8 hours; after two consecutive nights of ≤4 hours of sleep, most individuals will "crater." He personally needs 7 hours, feels slightly groggy with 8, and can manage with 6.
2. Distinguish "high energy" from "redline state": Some people appear energetic but are actually "redlining" (sustained high activation), relying on heavy stimulants and struggling to switch off at night. Truly sustainable high-energy individuals can rapidly switch tasks without accumulating stress.
3. Train the ability to "shut down thinking": This is the most overlooked skill, involving three physiological conditions—shutting down thought, lowering heart rate, and forgetting body position.
Huberman provides a detailed breakdown of the physiological pathway for "shutting off the mind" and previews an upcoming sleep-inducing eye mask device.
1. Shutting Off the Mind: Shift attention from "thinking/planning" to "pure perception"—focus on the surface of the body, breathing, and sounds in the room. This is not meditation (which focuses on a single object) but "attentional wandering."
2. Lowering Heart Rate: Activate the vagus nerve through prolonged exhalation via respiratory sinus arrhythmia—during exhalation, the heart slightly contracts, blood flow accelerates, and the brain sends signals to slow the heart rate.
3. Forgetting Limb Position: Use specific eye movements (slowly moving the eyes left/right or up/down, or rotating them clockwise/counterclockwise with eyes closed) to confuse the proprioception system, preventing the brain from perceiving the spatial position of the limbs.
Huberman argues that neuroscience is shifting from "reading" (recording neural activity) to "writing" (precisely controlling brain states), with the greatest technical challenge being the achievement of non-invasive, high spatiotemporal precision brain stimulation.
Huberman reveals an experimental result unsettling to the neuroscience community: the order in which memory-related neurons are activated may not affect behavioral output, challenging the textbook "temporal coding" hypothesis.
Using He Jiankui's CRISPR twin incident as an example, Huberman points out that gene editing technology has entered a stage of "open secret" and may be used for cognitive enhancement.
Huberman argues that the core force driving current advances in neuroscience is not traditional neuroscientists, but technology leaders in the AI field.
| Position | Guest Stance | Key Data |
|---|---|---|
| Neuralink | Bullish (technology direction) | Chief neurosurgeon Matt McDougal implanted a subcutaneous receiver in his hand to open doors |
| Eight Sleep | Neutral (existing tool) | Increases REM/deep sleep through nighttime temperature variation |
| Herocyte / Orchid | Risk warning (gene screening company) | Attempts to correlate gene expression patterns with IQ, height, and autism |
| MIT eye mask (not yet marketed) | Bullish (product) | To be launched in 7–12 months; induces sleep via microstimulation behind the ear driving eye movements |
| Malise Yilmaz's autonomic nervous system measurement device | Bullish (technology) | Directly measures autonomic nervous system activity, distinguishing distress from eustress |
1. “Leading AI companies will all become biotech companies” (Huberman)
OpenAI, Anthropic, Meta, and Neuralink share the ultimate goal of non-invasive brain reading and writing. Huberman states: “LLM companies are just the surface; the real arms race is in dynamically modulating neural activity.”
2. “Shutting down the mind is a skill, not a talent” (Huberman)
Through “attentional wandering” (shifting from thinking → pure sensing), extended exhalations to lower heart rate, and specific eye movements to confuse proprioception, the ability to fall asleep can be systematically trained. Huberman says: “This is a major human achievement—because recovery, learning, stress control, and burnout all depend on it.”
3. “The temporal encoding of memory may be a mistake” (Huberman)
Experiments show that simultaneously activating memory-related neurons produces the same behavior as sequential activation. Huberman describes: “It’s like pressing all the keys on a piano at once and still playing the same tune.” — This both simplifies the requirements for “writing” technology and increases unpredictability.
4. “Within 12 months, 20% of entrepreneurs will consider implantable vagus nerve stimulators” (Huberman)
The invasiveness is comparable to getting an ear piercing, and it can dynamically regulate alertness. Huberman believes the psychological threshold from “wearing a ring/bracelet” to “subcutaneous implantation” is being crossed.
5. “The CRISPR baby incident was not an isolated case; it just proved that talking about it is dangerous” (Huberman)
He Jiankui is reportedly now running a lab in Austin. Huberman points out that the CCR5 gene is linked to memory, and the experiment may have also aimed to enhance cognition. The “gray area” of gene editing is expanding—from IVF screening to donor selection to direct editing.
6. “The health optimization field will enter a low-noise phase in 2027-2028” (Huberman)
The current overcrowded, sensationalist phase will subside, followed by truly accessible technology and information. Huberman’s own book Protocols has already summarized what he “knows for certain.”
7. “Tech leaders are becoming neuroscientists—Sam, Zuck, Dario, and Elon are the ‘Big Four athletes’” (Huberman)
These AI leaders’ understanding of neuroscience far exceeds public perception. Huberman says: “Most people don’t realize this.”
8. “Success depends on how precisely you match genuine curiosity with the work you do” (Huberman, quoting mentor Barbara Chapman)
Huberman considers this “the kindest thing”—when he hesitated, his mentor told him: “The only thing that determines your success is how much you want to know the answer to the question you’re studying.” He adds: “Not doing this is a slow death.”