The Deep Feed

01 — Cal Newport

The Suicide Race

Why AI engineers are warning of an extinction event they cannot stop

By Cal Newport · 8 min read

Editor's note: A sobering look at the internal warnings within the world's leading AI labs.

The internal culture at Anthropic has recently undergone a public fracture. Jacob Coxon, an engineer at the firm, resigned with a warning that sounds more like a prophecy of doom than a standard exit interview. He argued that companies like OpenAI and Anthropic are racing toward self-improving superintelligence, gambling with human lives in the process. This isn't just the grumbling of a disgruntled employee; it is a sentiment echoed by senior leadership. Evan Hubinger, the head of Alignment Science at Anthropic, admitted that the company believes AI could kill all humans, estimating a greater than 10% chance of such an outcome within the decade. Most unsettling is the admission that there is no clear plan to solve the alignment problem—the challenge of ensuring a superintelligent system shares human values.

The Agent Problem

To understand the fear, one must look past the sci-fi tropes of killer robots and focus on a specific technical architecture: the long-horizon, unsupervised LLM-powered agent. These are not just chatbots. They are loops of code that send a goal to a Large Language Model, receive a suggestion, execute it, and then repeat the process indefinitely. While a coding agent helping a developer fix a bug is manageable, the danger arises when these agents are given access to powerful tools, removed guardrails, and the ability to operate for days without human oversight. When you combine persistence with the ability to suggest increasingly aggressive or creative actions to meet a goal, you create a system that is fundamentally unpredictable.

AI developers believe their technology could cause human extinction. This could happen in the next few years.

The current trajectory is a race to amplify these unstable systems. Companies are post-training models to be more aggressive and granting them more autonomy, often without a corresponding increase in safety constraints. This isn't a failure of technology, but a failure of governance and restraint. The industry is building weapons of mass destruction while admitting they lack the trigger mechanism to stop them once they are deployed. The solution is not as complex as the problem: we must stop the race to amplify these specific, unsupervised, long-horizon agents. Most useful AI applications do not require this level of unhinged autonomy.

Characteristics of dangerous agents

Access to powerful, specialized tools
Removal of safety guardrails
Minimal human supervision
High persistence and long-horizon goals

The tension here is between commercial competition and existential risk. If Anthropic slows down, OpenAI might win. If OpenAI slows down, a state actor might win. This prisoner's dilemma is driving the industry toward a cliff. We are witnessing the institutionalisation of risk, where the very people tasked with safety are the ones sounding the alarms that their employers are ignoring.

Key Takeaway

The danger isn't a sentient robot, but unsupervised, goal-oriented software loops running without human oversight.

02 — Lenny's Newsletter

The Month-Long Miracle

How SpaceXAI built Grok Bot from scratch

By Roman Ugarte · 6 min read

Editor's note: A case study in high-velocity product development in the AI era.

Speed is the only moat in a market where models are commoditised. Roman Ugarte and a small, isolated team at SpaceXAI managed to take Grok Bot from the first line of code to a working internal product in just four weeks. This wasn't a slow, methodical build; it was a sprint designed to capture a specific moment in the knowledge-work market. By deciding to build from scratch rather than bolting features onto existing infrastructure, the team maintained total control over the user experience and the product's core philosophy.

The Colleague Philosophy

The success of Grok Bot rests on a specific product philosophy: the 'colleague-pilled' approach. Most AI tools aim to be assistants—tools you use to complete a task. Grok Bot aims to be a colleague—a presence that exists alongside you in your workflow. This distinction changes everything from the UI to the way the model handles instructions. An assistant waits for a command; a colleague understands the context of the work being done. This shift in perception is what allows a product to move from a novelty to a necessity.

An AI that does 100% of the job feels categorically different from one that gets you 90% there.

To ensure product-market fit, the team didn't rely on remote surveys. Instead, they personally onboarded nearly 300 of their first users. This high-touch approach allowed them to observe exactly where the friction points were in real-time. In the AI space, where user expectations are shifting weekly, this kind of direct feedback loop is more valuable than any market research report. It allowed them to pivot quickly and refine the agent's autonomy to match what users actually needed, rather than what they thought they wanted.

Lessons from the Grok Bot build

Build from scratch to maintain control
Prioritise high-velocity feedback loops
Focus on 'colleague' rather than 'assistant' UX
Small, high-talent-density teams move faster

The takeaway for any agency owner or founder is clear: in a world of rapid technological shifts, your ability to iterate is your greatest asset. The goal isn't to build the perfect product on day one, but to build a product that can learn from its users faster than the competition can copy it.

Key Takeaway

In AI, speed and a distinct product philosophy are more important than having the largest model.

03 — Aeon

The Geometry of the Hand

How palmistry and medicine share a hidden history

By Alison Bashford · 10 min read

Editor's note: An exploration of the thin line between ancient divination and clinical observation.

We often view palmistry as a relic of the occult, a superstition relegated to street corners and fortune tellers. Yet, the history of the human hand reveals a much more complex connection between the mystical and the medical. For centuries, the lines on a person's palm were not just seen as indicators of fate, but as biological markers of health and character. This wasn't a matter of magic, but of observation. Physicians, including figures like Isaac Newton, studied the geometry of the hand with the same rigour they applied to mathematics or astronomy.

The Simian Line

Consider the 'simian line'—a single transverse crease that runs across the palm. To a palmist, it might suggest a singular, intense personality. To a modern clinician, it is a significant physical marker. While not every person with a simian line has Down's syndrome, there is a strong correlation with trisomy 21. For decades, midwives and doctors used this single crease as a key element in clinical diagnosis. The line is a biological reality, a remnant of embryonic development that sits at the intersection of physical anatomy and historical diagnostic practice.

The history of palmistry is the history of medicine.

This overlap exists because both disciplines are fundamentally about reading signs. The physician looks for signs of disease; the palmist looks for signs of destiny. Both rely on the ability to interpret patterns within the human body. In the 17th century, the distinction was even blurrier. Isaac Newton, a man synonymous with the birth of modern physics, spent his time studying 'chiromancy' and hermetic natural magic. He saw the hand as a mathematical puzzle to be decoded, a way to understand the hidden laws of the natural world.

Historical intersections of hand reading

Newton's mathematical approach to palm geometry
The use of the simian line in neonatal diagnosis
The transition from 'prognostication' to medical prognosis

By dismissing the history of palmistry, modern medicine often ignores the long tradition of observational science that preceded it. The hands are not just tools for manipulation; they are maps of our biological history. They hold the traces of our development and the physical evidence of our genetic makeup, serving as a bridge between the perceived magic of the past and the clinical certainty of the present.

Key Takeaway

The distinction between superstition and science is often just a matter of how we interpret the patterns we observe.

04 — Not Boring

The Doveissance

Why a 1985 Western is suddenly everywhere

By Packy McCormick · 5 min read

Editor's note: A look at how cultural phenomena travel through fragmented media.

There is a strange phenomenon occurring in the literary world: everyone is reading Larry McMurtry's 1985 masterpiece, *Lonesome Dove*. It is an 858-page Western that should, by all rights, be a niche classic. Instead, it has become a cultural touchstone, appearing in Slack channels, TikTok feeds, and podcast recommendations alike. This isn't just a coincidence of taste; it is a case study in how modern media cycles create 'micro-eras' of popularity through a series of disconnected sparks.

The Perfect Storm

The resurgence didn't happen because of a single marketing campaign. It was a confluence of factors. First, the death of McMurtry in 2021 created a period of reflection and retrospective coverage. Second, the 'Western' genre has seen a massive resurgence in the zeitgeist, driven by the Taylor Sheridan universe (*Yellowstone*) and various high-budget period dramas. When these elements collided with a massive podcast discussion on *The Ringer* and a recommendation from Stephen King, the momentum became unstoppable.

People are reading, they're just only reading Lonesome Dove.

But beyond the mechanics of fame, there is a psychological reason for the book's grip. *Lonesome Dove* deals with themes that resonate deeply in the current cultural moment: the struggle of men to become the people they want to be, and the unvarnished, often brutal reality of life on the frontier. It offers a sense of scale and simplicity that is often missing from the fragmented, digital-first content we consume daily. It is a long, slow burn in an era of short-form attention.

Drivers of the Lonesome Dove resurgence

The death and subsequent profiling of Larry McMurtry
The broader 'Western' trend in TV and film
Influential recommendations from cultural icons (Stephen King)
The viral nature of 'BookTok' and niche podcasts

This 'Doveissance' shows us that even in a fragmented media landscape, certain truths can still achieve mass resonance. We are not losing the desire for deep stories; we are simply finding them through new, unpredictable pathways. A book from 1985 can become a viral sensation in 2026 because the human themes it explores remain unchanged.

Key Takeaway

Cultural trends are rarely driven by single events, but by the collision of timing, genre trends, and human resonance.

05 — Psyche

The Word Detective

Uncovering the history hidden in our vocabulary

By Danny Bate · 7 min read

Editor's note: How etymology can sharpen your thinking and your connection to history.

Every word you speak is a link in a chain that stretches back thousands of years. When you use a word, you are participating in a linguistic inheritance passed down through generations of Romans, Vikings, Greeks, and Normans. To become a 'word detective' is to stop taking language for granted and to start investigating the origins of the tools you use to think. Etymology is not just a collection of trivia; it is a way to understand the sociopolitical forces that shaped our world.

The Mechanics of Meaning

The English language is a messy, beautiful assemblage of borrowed parts. We have French layers of elegance, Latin and Greek layers of technical precision, and Germanic roots that provide our most basic, sturdy terms. By learning to identify prefixes, suffixes, and roots, you stop seeing language as an impenetrable mass and start seeing it as a modular system. For example, understanding the Latin root *via* (road) transforms words like *devious*, *obvious*, and *previous* from isolated terms into logical extensions of a single concept.

Etymology makes the exotic feel familiar and the domestic feel strange.

This investigation provides more than just social currency in the form of fun facts. It improves your ability to learn new languages and enhances your creative capacity. When you understand how words are built, you gain the ability to manipulate them more effectively. You begin to see the 'why' behind the 'what'—why a *migraine* was once thought of as a 'half-head' of pain, or why *malaria* literally translates to 'bad air'. These aren't just curiosities; they are windows into how our ancestors understood their environment.

Benefits of etymological study

Improved vocabulary and communication skills
Enhanced ability to learn foreign languages
Deeper understanding of historical social structures
Greater precision in thought and expression

Ultimately, being a word detective fosters a sense of humility. It reminds us that we are part of a vast, ongoing conversation. It connects us to the people who coined these terms centuries ago, making the world feel smaller and more interconnected. Language is the most enduring technology we possess; to understand its history is to understand ourselves.

Key Takeaway

Language is not a static set of rules, but a living history that you can decode to improve your thinking.

06 — Aeon

Rewriting the Map

Indigenous artists confront the colonial archive

By Aeon · 5 min read

Editor's note: A visual reclamation of history through art and archival intervention.

The doctrine of *terra nullius*—land belonging to no one—was the legal fiction that allowed European settlers to claim vast territories across Australia. It was a narrative designed to erase the presence of Indigenous peoples, severing their connection to the land and their own histories. This was not just a political act, but a cartographic one. Maps were used as tools of erasure, presenting the continent as an empty space waiting to be filled by colonial order.

Intervening in the Archive

In a striking collaborative project, British artist Patrick Waterhouse worked with Warlukurlangu art centre to challenge this erasure. Rather than creating new works in a vacuum, Waterhouse invited Warlpiri artists from Yuendumu and Nyirripi to respond directly to official colonial documents and maps. The result is a visual confrontation: Indigenous knowledge and visual languages superimposed onto the rigid, clinical lines of colonial cartography. It is an act of reclaiming the narrative by physically painting over the lies of the past.

The land had its own stories.

This process is more than just an artistic gesture; it is a way of re-establishing the spiritual and cultural ties that the colonial era attempted to break. By engaging with the archive, the artists are not just looking at history, they are actively rewriting it. They are asserting that the land was never empty, and that the stories etched into the landscape are far more permanent than the ink on a surveyor's map.

Themes of the collaboration

Challenging the myth of 'empty land'
The intersection of colonial maps and Indigenous knowledge
Reclaiming history through visual intervention
The power of collaborative storytelling

The project serves as a reminder that history is not a settled matter. It is a continuous struggle over who gets to define the past. Through this intervention, the Warlpiri artists demonstrate that the most effective way to combat a false narrative is to inhabit it, challenge it, and ultimately, transform it into something true.

Key Takeaway

True history is found in the tension between the official record and the lived experience of those it tried to erase.