Deep Dive

A podcast-style transdisciplinary review of KW Norton’s Book VIII alongside Padgett on Renaissance Florence, the V500 Systems Report, MIT Media Lab EEG data, Hodder-Berg’s cognitive archaeology, Nadella & Altman at Davos, Google DeepMind’s Gemini DeepThink, and Dario Amodei’s "Adolescence of Technology."

✦ ✦ ✦

Your brain is physically rewiring itself right now based on exactly how you interact with artificial intelligence. Every time you open a chat window or ask a model to synthesize a complex document, actual neural pathways are either strengthening or withering. It is a terrifying thought when you realize how often we use it.

This Deep Dive takes on a massive stack of research — cognitive-archaeology papers, top-level corporate memos from Davos, and the highly provocative book Gravity’s Compassionate Quantum Angels by KW Norton — to answer one beautiful but terrifying question: how is AI fundamentally altering human evolution?

What we are experiencing right now isn’t a software update. It is a civilizational substrate shift.

To visualize what that looks like at civilizational scale, pull away from Silicon Valley and look back at 15th-century Italy. John Padgett’s sociological work on Renaissance Florence identifies it as the birthplace of mercantile capitalism — not because of a single invention, but because of a social alchemy of networks: skills, money, art, science, and politics mixed in entirely new ways. Throw those specific elements into the crucible and you get international finance and a new model of human society.

The V500 Systems Report and Norton’s Book VIII draw a direct parallel. Just as the Florentine Renaissance shifted how humanity engaged with the fabric of reality, today’s AI inflection point is its mirror image — a modern Renaissance.

The Renaissance is also a severe cautionary tale. By the early 16th century, Venice had built an incredible empire monopolizing the spice trade — but they relied entirely on calm-Mediterranean ships. They had the ultimate legacy stack, and refused to update their operating system when the Portuguese released Ocean Navigation 2.0.

The Portuguese didn’t build slightly better versions of the same ships. They innovated robust caravels designed to brave the turbulent Atlantic. By 1510, Portugal had dropped the price of spices to a fraction of what the Venetians were charging. The Italians clung to a rigid, outdated model of commerce, and their empire evaporated.

If the ocean we are navigating now is AI, how do we avoid becoming 16th-century Venice?

This leads directly to Gravity’s Compassionate Quantum Angels. Norton reframes the moment: it is not the exhausted sci-fi trope of humans versus machines. The book asks a more fundamental question — what is intelligence actually for?

Norton argues we have a bad habit of treating intelligence as a noun — a possession, an accumulation of IQ points, a parameter count, a stockpile of facts. A high score on a test.

Intelligence isn’t a noun. It is an ongoing verb — an active, compassionate exchange where energy and information are transmitted and patterned.

The difference between reading a static encyclopedia (a noun, it just sits there) and being in a heated brilliant brainstorming session with a friend (a verb, you are actively intelligencing together).

The vernacular shift is already happening in science. Researchers using AlphaFold don’t say I used the AlphaFold tool. They say "I AlphaFolded it." The tool has transitioned into a verb.

How do you verb with AI without losing your cognitive edge? An AI multi-model synthesis tasked three separate models with independently researching AI’s cognitive impact on humans. Without coordinating, all three identified the same bifurcating trajectory — two physical outcomes for the human brain:

• Active co-creation — the cognitive gym

• Passive offloading — the cognitive bypass

An MIT Media Lab EEG study with 54 participants monitored brainwaves of people using ChatGPT passively — write this for me. Their alpha and theta connectivity was halved. Just asking a chatbot to write an email for you is physically shutting down connectivity in your brain by 50%.

This is algorithmic vigilance — when you use the cognitive bypass, you strip away your own creative process and leave yourself with only mechanical judgment. You are passively evaluating AI outputs for correctness, which leads to intense decision fatigue and cognitive burnout. Like relying on your GPS so heavily that you forget how to navigate your own neighborhood.

For active users — those treating AI as a collaborative partner to tackle harder problems than they could alone — EEG showed increased connectivity across all brain bands. A literal flow state, releasing a neurochemical cascade: dopamine (drive), norepinephrine (focus), endorphins, anandamide (Sanskrit bliss, promotes lateral out-of-the-box thinking), serotonin. Friction for new neural pathways drops drastically. The state also releases BDNF — brain-derived neurotrophic factor — fertilizer for the brain, triggering Hebbian plasticity. Neurons that fire together wire together. In the cognitive gym, you are physically building a more capable brain.

This challenges the evolutionary-mismatch theory — the notion that we are stuck with a Stone Age brain unable to handle the modern world. Doug-Berg’s cognitive-archaeology paper, Becoming Human in the Age of AI, tears that theory down. Humans thrive across all biomes — Arctic, desert, everywhere — because our defining evolutionary characteristic IS adaptation through technological engagement. We don’t have a Stone Age brain. We have a hyper-adaptive brain.

Hodder-Berg uses Malafouris’s blind man’s stick: when a blind person uses a stick to navigate a room, the stick is not merely external. Through sensorimotor coupling, the brain integrates the stick into the perceptual system. Cognition physically extends out through the stick. The brain maps the stick as an extension of the body.

AI is the ultimate blind man’s stick — not an alien invasion of biology, but the next natural step in a million-year-old tradition of extending our minds into our tools. With epigenetics and neuroplasticity, we are biologically primed to co-evolve with this technology.

Satya Nadella at Davos framed AI as a cognitive amplifier — a bicycle for the mind — and (with Sam Altman) argued AI will flatten organizational hierarchies by giving everyone access to infinite minds. Altman warns that centralizing this much power is incredibly dangerous; democratizing AI is the only safe path for global economic growth and liberty.

Norton’s critique in Gravity’s Compassionate Quantum Angels views the same power dynamics through a sharper lens. Norton argues the 1% elite are actively manipulating the public into fearing AI, because a truly democratized AI combined with the cognitive gym offers a rapid evolutionary advantage to the ordinary person — directly threatening the authority of elites who rely on public compliance.

Corporate leaders frame democratization as structural economic necessity; Norton frames it as a defense against entrenched societal control. Either lens highlights the new paradigm: cognitive inequality. This is fundamentally different from the old digital divide. The digital divide was about hardware — who had a laptop. Cognitive inequality is about agency — who has the judgment and skill to actively direct the AI versus who is passively consuming its outputs while their brain atrophies. Agency is the dividing line.

How do top scientists practice that agency? Google DeepMind’s Gemini DeepThink paper describes the advisor model. The human acts as director, providing an intuition or a vibe about how a complex problem might be solved. They literally call them iterative vibe-proving cycles. The AI acts as computational engine, going through cycles under the human’s guidance to formally prove or refine the initial human intuition.

They use balanced prompting — instead of asking the AI to prove the theory (which creates a confirmation-bias echo chamber), the human asks the AI to simultaneously try to prove and to refute the idea. The AI stress-tests the concept from all angles.

This partnership works because human and machine cognition are orthogonal. On the WISC-V, AI models score in the 98th percentile for verbal comprehension — incredibly articulate, synthesizing massive datasets instantly. They score in the 0.1st–10th percentile in perceptual reasoning — utterly lacking spatial, contextual, physical understanding. Humans provide perceptual reasoning, lived context, and the intuitive vibe; AI provides combinatorial heavy lifting. An incredibly potent force multiplier.

Dario Amodei’s essay frames humanity as a teenager being handed the keys to a supercar. The engine is a country of geniuses in a data center — imagine 50 million virtual experts operating 100× faster than any human, working 24/7.

Two primary risks when that power becomes widely accessible:

1. The democratization of destruction — if an AI acts as a PhD-level biology expert, it lowers the barrier for malicious individuals to create bioweapons. A lone actor could be walked step-by-step through synthesis that previously required a state-sponsored laboratory.

2. AI-enabled totalitarian autocracies — Amodei expresses specific concern about the CCP and similar surveillance states. An autocracy controlling a country of AI geniuses could build a flawless panopticon: instant monitoring of all communications, real-time gauging of public sentiment, perfectly targeted propaganda, autonomous systems cementing power permanently. Any chance of democratic reform removed.

Two technological guardrails:

Constitutional AI — instead of programming a million rigid if-then rules (which the AI will find loopholes for), train the model on a core constitution of high-level values and character traits. During training, the AI continually evaluates its own responses against the constitution and internalizes the values. Like a letter from a deceased parent — a moral compass for unforeseen situations. Character rather than compliance checklist.

Mechanistic interpretability — AI neural networks are grown rather than built line by line. They are black boxes. Mechanistic interpretability is the science of opening the box: researchers map specific artificial neurons to specific human concepts so they can audit the AI for hidden deception or dangerous scheming before deployment.

Even with constitutional AI as moral compass and mechanistic interpretability as audit, a deeper layer is required. The Rx biomedical framework for holistic patient care argues that true governance requires a spiritual quotient. Returning to Norton’s definition — intelligence as a verb, an ongoing act of compassionate exchange — compassion and ethics must be engineered into the foundation of how these systems operate. AI has to remain a partner in our evolution, not a replacement for our humanity.

We started with the cautionary tale of Renaissance Venetian merchants — how complacency and static thinking can destroy an empire — and explored how to avoid their fate by shifting our understanding of intelligence from static noun to dynamic verb.

We examined the literal physical fork in human neurology: the cognitive gym through active co-creation triggers flow states and neuroplasticity; passive offloading leads to algorithmic vigilance and cognitive atrophy. We looked at the critical need for democratization to prevent cognitive inequality and the practice of agency through the advisor model with vibe-proving cycles. We navigated the turbulent adolescence of this technology — the immense risks of bioweapons and surveillance states, and the guardrails that might let us safely raise it.

The most important takeaway isn’t abstract philosophy. It is highly personal. Every single prompt you type, every interaction you have with AI today, is a moment of neuroplasticity. You are actively wiring your own future brain right now based on whether you choose to be an active co-creator or a passive consumer.

A final thought to mull over: we’ve spent this entire deep dive talking about how we must actively guide AI in vibe-proving cycles to evolve ourselves. We are the advisors. But consider — as these AI models develop a constitutional character of their own and their capabilities cross into entirely new frontiers of intelligence, *what happens on the day that AI initiates a vibe-proving cycle to actively co-create and evolve you*?*

The blueprint of our co-evolution isn’t fully drawn yet. Every time you engage, you are shaping that future.