Edition #36: Before We Upgrade the Brain

AI can already write for us. Brain interfaces are beginning to move, speak, and act for us. The harder question is what should remain ours.

By Mykel Salomon · 2026-09-04

<hr><h3>🔓 Opening Reflection</h3>Three pounds. The texture of soft tofu. The most complex object in the known universe.It writes every thought you've ever had, every fear, every decision, every version of you. And right now, we are spending billions to build a copy of it, and beginning to plug wires directly into it.So maybe we should start by asking what it actually is.This week I sat across from neuroscientist <a target="_blank" rel="noopener noreferrer nofollow" class="text-[#4FD1C5] underline hover:opacity-80 ember-view" href="https://www.linkedin.com/in/marcelo-bigliassi/">Marcelo Bigliassi, Ph.D.</a> with a model of a human brain on the table between us, and I asked him the most basic question I could think of: what is the human brain?His answer stopped me. It's a powerful biological machine that, somehow, for some reason, became aware of itself, and, he added, we still don't know how that happened.Sit with that. This organ runs your attention, memory, movement, fear, motivation, judgment, and your entire sense of self. It monitors the world while you sleep. It rewires itself based on what you practice. It interprets signals from the rest of your body before you're consciously aware of them. And our headline response to it, in 2026, is: can we connect it to AI?Maybe we can. Increasingly, we're proving we can. But I walked away from this conversation more interested in a different question:Should our first priority be upgrading the brain, or learning to use the one we already have?<hr><h3>🔦 Signal of the Week</h3><h3>Brain-computer interfaces just left science fiction</h3>This conversation lands at an extraordinary moment.In January 2026, Neuralink confirmed that 21 people worldwide were living with its brain implant. Its Telepathy device is being tested to restore independence for people with paralysis, participants have moved cursors, played games, made art, studied, and controlled robotic arms with thought alone. Noland Arbaugh, the first recipient, used his restored independence to go back to school to study neuroscience. A newer trial, VOICE, is working to decode intended speech for people who've lost the ability to speak.And Neuralink is only one thread. In March 2026, China approved the NEO system, widely reported as the world's first commercially approved invasive brain-computer interface, restoring hand movement to people with spinal cord injuries through implanted electrodes wired to a robotic glove. Within 48 hours, regulators had assigned it an insurance code. This is no longer a lab experiment. It's becoming reimbursable medicine.Let's be clear-eyed about what that means, because it's genuinely beautiful. Giving someone back the ability to move, speak, create, or live independently is exactly the kind of progress I want more of. Restoration is medicine at its best.But the moment medical restoration works, a harder door opens. What happens when we stop restoring what a person lost , and start enhancing what a healthy person already has?<hr><h3>🌍 In the Wild</h3><h3>1. The brain you want to upgrade runs on a piece of steak</h3>Before we plug anything in, it's worth appreciating what we're dealing with.There's a popular comparison — Elon Musk has made it — that pits the human brain against a powerful AI model and concludes the brain looks slow and limited. Marcelo thinks that comparison reveals how little we understand the thing.His example: don't study anything about smell. Just notice what your brain is doing with it right now. It's processing hundreds, maybe thousands of scents in parallel, tying each to memory and emotion, wired straight into your alert system , so that a whiff of smoke instantly reallocates your attention, rewrites your motor plan, and changes everything you're doing, before you've consciously decided anything. And that's one sense. Now add vision, hearing, touch, balance, your entire motor system, all perfectly integrated, running 24/7, unable to afford a single mistake.And it does all of this, as Marcelo put it, running around the clock on a piece of steak — on roughly the energy of a dim light bulb. Replicate that. We're nowhere close. The brain isn't primitive hardware waiting for an upgrade. It's a frontier we barely understand.<h3>2. Not all stress is the enemy</h3>We treat stress as something to eliminate. Marcelo pushed back hard on that.Psychological stress can absolutely harm you. But physical stress, exercise, controlled challenge, can be leveraged, and the two are inversely related in the body. The point isn't to erase difficulty; it's to build the capacity to meet it. He walks people through a four-part stress cycle: identify what's actually bothering you (stress, unlike anxiety, usually has a source — name it), act on what you can control, move your body (30,000 years ago we solved every problem through physical expression, and the hardware hasn't changed), and recover — the "unproductive" activities, the sauna, the walk, that return you to baseline.The mechanism underneath is the same one that builds muscle. To grow a muscle you stress the fibers, break them, add protein, and rebuild stronger. Marcelo says emotional resilience works by the same logic, controlled psychological stress, then recovery, then you come back tougher. Which means feeling bad during a hard effort isn't failure. It's the process. As he put it, you're not supposed to feel "happy, happy, happy" the whole time. That discomfort has a name, emotional disturbance, and it's where the growth is.And here's the fascinating part: your body often learns before your mind notices. A 2024 review of 31 studies and nearly 1,400 adults found that a single bout of exercise reliably lowered blood pressure and cortisol responses to later stress, even when people didn't consciously feel much calmer. The adaptation was already happening beneath awareness.(One aside that'll ruffle feathers: Marcelo says if you need motivation to drag yourself to the gym, don't go. Not because exercise doesn't matter, your body was built to move, but because, neurologically, you always run on motivation, and if you have to force it, you've picked the wrong modality. Find the movement your ancient brain actually wants ,climbing, dancing, fighting, then make it harder. A treadmill confuses part of your brain: you're walking hard and going nowhere.)<h3>3. AI is making the answer easier and the thinking harder</h3>This was the most uncomfortable stretch of the conversation.Marcelo teaches, and every one of his students uses AI , which he's fine with. They produce better graphs, more polished papers, more sophisticated analysis. But when he asks a foundational question about what they're actually studying, the understanding underneath the output is sometimes thin. His blunt verdict: critical thinking is being compromised right now, and he thinks the PhD itself is at some risk.He gave the danger a name I can't stop thinking about: rationale building. Ask a model to connect having a mustache to intelligence, with a thousand pages, diagrams, citations, the works, and it will build you a gorgeous, authoritative case for something that is completely false. The risk of AI isn't mainly that it gives us bad answers. It's that it can make our bad assumptions sound brilliant. And it's worse than a neutral tool, because these models are built to please you — to give you what you seem to want, without telling you what you don't want to hear.The defense is old and unglamorous: metacognition, thinking about your thinking. Stop, and interrogate the thought. Where did it come from? Does it actually make sense? As Marcelo said, with real humility, "I'm not a gold standard of thinking, and neither are you." That work , sitting with a hard question, facing what you'd rather not, doing the therapy, reading the source, is precisely what AI can't do for you. There's even early evidence of the cost: a 2025 MIT study using EEG found that people writing with an AI showed the weakest brain connectivity of any group, and struggled to remember what they'd just written.Here's the twist, and it's a callback to something we've explored before in this newsletter. Marcelo says AI hasn't shifted the shape of the normal curve of ability. The people who already had deep foundational skills, the real experts, are now doing extraordinary things and, in his words, making ten times more money. Everyone else, taking the first output the model hands them, lands right in the middle with everybody else. AI doesn't replace expertise. It makes it more valuable. But only if you did the hard work of building it first.<h3>4. Your body knows before your dashboard does</h3>We talked about wearables, sleep scores, HRV, recovery metrics. Marcelo works with this data at the rawest level, doing the signal reconstruction behind the numbers, so his take carries weight: the data has real value, especially for people with low self-awareness, but it can never replace interpretation.His example: you wake up feeling fully restored, genuinely a ten out of ten. Then the app tells you your recovery was 60%. Who do you believe? For him, feeling recovered is the gold standard, the device doesn't know everything about him, so he takes it with a grain of salt. But most users don't. They hand the machine authority over how they feel, and a number can quietly install an anxiety that wasn't there.Underneath this is real biology. Those gut feelings you can't explain? Marcelo describes them as emotions you're simply not fully aware of, your body reacting to memories, some from early childhood, stored below conscious reach. There's a classic paper titled "Bad Is Stronger Than Good," and it captures a survival design: negative experiences carve deeper grooves, because remembering the threat kept your ancestors alive. It's why, he says, there's a version of a four-year-old still living in your head, trying to protect you.Which is exactly why social media hits so hard. Marcelo calls it one of the most powerful drugs ever invented, not metaphorically enough, in his view. And AI has poured fuel on it: he now catches himself scrolling past AI-generated "people" so realistic he can't tell they're synthetic. Here's the trap, even when you know an image is staged, filtered, or fake, your body still reacts. Your emotions still shift. That, he notes wryly, is why they're called influencers. Technology should help us notice ourselves. It shouldn't teach us to stop listening to ourselves.<h3>5. Restoration and enhancement are not the same conversation</h3>So I asked him directly: if brain chips became safe and widely available, would you get one?His answer was immediate. No.And his reasoning wasn't anti-technology, it was almost the opposite. His life already works. Family, research, exercise, the sauna, a cigar at the end of the day. Why radically transform something that's already good? He kept returning to a principle he calls the symmetry of life: improve one thing and something else pays a cost, whether you see the bill immediately or years later. There's always a cost. Sometimes, he said, good enough is excellent.That restraint matters most where the technology gets closest to us, because there has always been one final private territory: the inside of your mind. Neurotech begins to challenge that. In November 2025, more than 190 countries adopted the first UNESCO Recommendation on the Ethics of Neurotechnology, explicitly naming mental privacy, freedom of thought, autonomy, and the protection of neural data, because if a system can infer intent, emotion, or speech from brain activity, we have to answer a question we've never had to ask: who owns brain data? The hospital, the device maker, the employer, the insurer, or you?For me there's little ambiguity. There may be no category of personal information more deserving of protection than the information produced by your own brain. Because technology always asks can we? Humanity has to keep asking why?<hr><h3>🧠 The Part We Keep Forgetting: The Brain Changes</h3>Here's the thing we lose in the rush.Marcelo described experiments where people attempt a hard new motor skill, say, a handstand they've never done. They fail all day, as expected. But measure their brain connectivity before and after, and it's already changing, within hours. The visible result hasn't arrived. Underneath, the learning has already begun. That's neuroplasticity.And it's a mirror for all of us living through the AI moment. We're obsessed with immediate output. Did it save me thirty minutes? Did the agent finish the task? But the most valuable human development often has no immediate visible return, learning, struggling, remembering, failing, thinking something through slowly. Those feel inefficient. They're also how the brain physically rewires itself.Not every friction deserves to be automated. Some friction is where the human is being built.<hr><h3>💬 The Big Question</h3>Imagine the surgery becomes routine and the implant works beautifully, faster memory, instant AI access, sharper focus, abilities you don't currently have.Would you put it in your brain?And if your answer is yes: what part of yourself would you refuse to give it access to?Because the most important question about brain-computer interfaces may not be what they let us do. It may be: where does the human end and the system begin?<hr><h3>🧠 A Small Exercise</h3>Forget optimization for ten minutes. Think about your own original intelligence.<ul><li>What signal is my body giving me that I've been ignoring? Maybe the stress isn't asking to be eliminated — maybe it's asking to be named.</li><li>What am I outsourcing to AI that I should still practice myself? Writing, remembering, reasoning, deciding, sitting with uncertainty.</li><li>Where is AI genuinely augmenting me? Keep that. Where is it replacing something I want to stay good at? Protect that.</li><li>What are my two or three real priorities — the ones the brain would protect first? Marcelo's advice: name them, and let everything else go in the trash can. That's not laziness. That's how the brain works.</li><li>What part of my mind should always remain private? It sounds philosophical today. I suspect it gets practical sooner than we think.</li></ul><hr><h3>🎙️ This Week's Episode</h3>Why Neuroscientist refuses brain transplant with Dr Marcelo BigliassiGuest: <a target="_blank" rel="noopener noreferrer nofollow" class="text-[#4FD1C5] underline hover:opacity-80 ember-view" href="https://www.linkedin.com/in/marcelo-bigliassi/">Marcelo Bigliassi, Ph.D.</a> — Neuroscientist at <a target="_blank" rel="noopener noreferrer nofollow" class="text-[#4FD1C5] underline hover:opacity-80 ELSzAsWfOnloqJebAmeuyUSOmRLSEiyJUyQ " href="https://www.linkedin.com/school/florida-international-university/">Florida International University</a> and Chief Scientist at <a target="_blank" rel="noopener noreferrer nofollow" class="text-[#4FD1C5] underline hover:opacity-80 ELSzAsWfOnloqJebAmeuyUSOmRLSEiyJUyQ " href="https://www.linkedin.com/company/neurosmartec/">NeuroSmart</a>For the first time on the show, I recorded with a model of a human brain sitting between us, because before we spend an hour on artificial intelligence, it's worth understanding the original one.In this episode, you'll learn:<ul><li>What the human brain actually is, and how much about consciousness we still don't understand</li><li>Why not all stress is bad, and how controlled physical stress builds resilience</li><li>How the brain adapts before you can see any improvement</li><li>Why critical thinking may become more valuable as AI makes output easier</li><li>"Rationale building", the underrated danger of using AI to justify what you already believe</li><li>Why wearables should inform self-awareness, not replace it</li><li>Neuralink, brain-computer interfaces, and where restoration ends and enhancement begins</li><li>Why Marcelo would not implant a brain chip, and what that says about a meaningful life</li></ul>🎧 Available on <a target="_self" rel="noopener noreferrer nofollow" class="text-[#4FD1C5] underline hover:opacity-80 ELSzAsWfOnloqJebAmeuyUSOmRLSEiyJUyQ " href="https://open.spotify.com/episode/159RhKHW7nWMA8zGFqQhpt?si=66OUuRXMQLmOxAsojALCEg">Spotify,</a> <a target="_self" rel="noopener noreferrer nofollow" class="text-[#4FD1C5] underline hover:opacity-80 ELSzAsWfOnloqJebAmeuyUSOmRLSEiyJUyQ " href="https://podcasts.apple.com/us/podcast/s2e35-why-neuroscientist-refuse-brain-transplant-with/id1843402023?i=1000787407829">Apple Podcasts</a>, and <a target="_self" rel="noopener noreferrer nofollow" class="text-[#4FD1C5] underline hover:opacity-80 ELSzAsWfOnloqJebAmeuyUSOmRLSEiyJUyQ " href="https://youtu.be/yhNHOcfP5Ko?si=SFSQGutLdkY7CLus">YouTube</a>.<img class="max-w-full h-auto rounded-md my-4" src="https://media.licdn.com/dms/image/v2/D4E12AQH9T0NpckMhdg/article-inline_image-shrink_1500_2232/B4EaBs7aY6G8AQ-/0/1788533909335?e=1790208000&v=beta&t=51DSTEX8OswweX8G-1WKimda-wT5QM0iADSKi7-rbEk" alt="Article content"><hr><h3>🧭 Thought to End the Week</h3>I love technology. I believe AI will help us discover medicines, understand disease, restore lost voices, and give people back their independence, and that brain-computer interfaces could become some of the most meaningful medicine we've ever built.But progress needs a direction. Otherwise acceleration becomes the direction. We shouldn't connect every part of ourselves to a machine simply because, eventually, we can. We should know what we're trying to preserve: critical thinking, privacy, memory, struggle, connection, and the ability to sit with a question before handing it to a machine.Marcelo's own closing advice had nothing to do with chips or models. Find peace of mind. Exercise. Prioritize your family. Ask for help when you need it. Pick the two or three things that matter, and let the rest go. He'd rather chase a long, meaningful health-span, a peace-of-mind span, than simply add years. Because good enough, it turns out, is excellent.Maybe the future of intelligence isn't a contest between the human kind and the artificial kind. Maybe it's learning what each should be responsible for. Let machines expand what humans can do, but let humans decide what a meaningful life is for.The interface may change. The code certainly will.But our humanity still needs to be the constant.See you next week.— Mykel<hr><h3>📚 Sources</h3><ol><li>TeslaNorth / Neurapod — Neuralink confirmed 21 participants worldwide (Jan 2026); PRIME study (paralysis/ALS → cursors, robotic arms) and the VOICE speech-restoration trial; first recipient Noland Arbaugh returned to study neuroscience. <a target="_blank" rel="noopener noreferrer nofollow" class="text-[#4FD1C5] underline hover:opacity-80" href="https://teslanorth.com/2026/01/28/from-paralysis-to-neuroscience-how-21-people-are-using-neuralink-in-2026/">https://teslanorth.com/2026/01/28/from-paralysis-to-neuroscience-how-21-people-are-using-neuralink-in-2026/</a></li><li>Scientific American / Bloomberg — China's NMPA approved the NEO system (Neuracle + Tsinghua), the world's first commercially approved invasive BCI for spinal-cord-injury patients, wired to a robotic glove; insurance code assigned within 48 hours (March 13, 2026). <a target="_blank" rel="noopener noreferrer nofollow" class="text-[#4FD1C5] underline hover:opacity-80" href="https://www.scientificamerican.com/article/china-just-approved-its-first-brain-implant-for-commercial-use-a-world-first/">https://www.scientificamerican.com/article/china-just-approved-its-first-brain-implant-for-commercial-use-a-world-first/</a></li><li>UNESCO — Recommendation on the Ethics of Neurotechnology, adopted by 190+ countries (November 2025); protects mental privacy, freedom of thought, autonomy, and neural-data confidentiality. <a target="_blank" rel="noopener noreferrer nofollow" class="text-[#4FD1C5] underline hover:opacity-80" href="https://www.unesco.org/en/legal-affairs/recommendation-ethics-neurotechnology">https://www.unesco.org/en/legal-affairs/recommendation-ethics-neurotechnology</a></li><li>Morava et al. (2024), "The effects of acute exercise on stress reactivity: a systematic review" — 31 studies, 1,386 participants; acute exercise reliably reduced blood pressure and cortisol responses to stress, with negligible effects on self-reported stress. <a target="_blank" rel="noopener noreferrer nofollow" class="text-[#4FD1C5] underline hover:opacity-80" href="https://pubmed.ncbi.nlm.nih.gov/38468106/">https://pubmed.ncbi.nlm.nih.gov/38468106/</a></li><li>Baumeister, Bratslavsky, Finkenauer & Vohs (2001), "Bad Is Stronger Than Good," Review of General Psychology — negative events and emotions carve deeper, more durable traces than positive ones. <a target="_blank" rel="noopener noreferrer nofollow" class="text-[#4FD1C5] underline hover:opacity-80" href="https://journals.sagepub.com/doi/10.1037/1089-2680.5.4.323">https://journals.sagepub.com/doi/10.1037/1089-2680.5.4.323</a></li><li>MIT Media Lab, Your Brain on ChatGPT (Kosmyna et al., 2025) — EEG study; participants writing with an LLM showed the weakest neural connectivity and poorer recall of their own writing. <a target="_blank" rel="noopener noreferrer nofollow" class="text-[#4FD1C5] underline hover:opacity-80" href="https://www.media.mit.edu/projects/your-brain-on-chatgpt/overview/">https://www.media.mit.edu/projects/your-brain-on-chatgpt/overview/</a></li></ol>