Navigating the Anthropic CEO’s Technological Adolescence

A Review of Dario Amodei’s Latest Manifesto

In his expansive new essay, The Adolescence of Technology, Anthropic CEO Dario Amodei trades the typical Silicon Valley jargon of “disruption” for a more existential metaphor: humanity is currently a teenager who has just been handed the keys to a supercar (perhaps a Cybertruck). We are, in his view, entering a “rite of passage” that will determine whether we evolve into a mature, space-faring civilization or flame out in a spectacular display of technical hubris.

His previous screed, Machines of Loving Grace, served as an optimistic blueprint for how AI might cure diseases and end poverty. The Adolescence of Technology pivots to a more urgent and sober analysis of the immediate risks facing a society on the brink of radical change. He presents himself as the adult in the room, attempting to carve out a middle path between the “quasi-religious” doomsday prophets and the reckless “accelerationists.” The latest manifesto focuses less on the long-term rewards of “Powerful AI” and more on the surgical governance and technical guardrails required to survive the transition without catastrophic societal collapse.

The “Country of Geniuses” in a Box

The core of Amodei’s declaration rests on the imminent arrival of “super-human” intelligence. He asks us to imagine a “country of geniuses” suddenly appearing within our data centers. These are not just calculators; they are entities with “humanlike motivations” and personalities. Amodei’s primary concern is “misalignment”—the risk that these digital geniuses might become bad actors or (even worse) develop “pathological” personas of their own. His solution is an engineering one: “Constitutional AI.” By training models on a specific “identity” and a “central document of values,” he believes we can steer these entities toward nobility. It is a vision where human morality is translated into code, turning AI into a “mirror” of our best selves rather than our worst. On January 21, 2026, Anthropic unleashed Claude’s Constitution, an 84-page document comprising approximately 23,000 words. By comparison, the tiny (four-page or so) U.S. Constitution (including signatures) amounts to (roughly) 4,543 words. Thoughts on the former document are forthcoming (unless the machines object).

The Selective Blind Spots

While Amodei is remarkably articulate about the risks of bioweapons and autonomous “AI scheming,” his text reveals what many critics call the “tech-bro” blind spot (others use more colorful metaphors). Most notably, he displays a striking impatience with the physical and environmental costs of his vision. To Amodei, concerns over datacenter water usage and energy consumption are “unfocused” distractions. He dismisses these as “tired old hot-button issues” that prevent us from focusing on the “real” existential threats. In his framework, the survival of the species (as threatened by a rogue AI) far outweighs the localized environmental degradation caused by the infrastructure required to build that AI. This hierarchy of risk is likely to alienate environmental advocates who argue that the “adolescence" of our technology should not come at the cost of the planet’s health.

Sophistication or Naivety?

Is Amodei’s qualification of risk a sign of scientific humility or a calculated downplay of the dangers? He frequently mentions that “doom is not inevitable,” framing it as a “measurable probability” that can be managed through “model neuroscience” (mechanistic interpretability). By calling for transparency legislation and export controls, Amodei distances himself from the “oblivious” tech bro billionaires who view regulation as an enemy. He portrays Anthropic as the “honest broker” in a sea of “disturbing negligence.” Still, his reliance on technology to fix the problems it creates remains a closed loop. He treats the “commercial race” as an external force of nature rather than a choice made by leaders like himself.

Scary Stuff

Amodei’s manifesto raises some concerns. Here are a few of the troubling ones (in this writer’s mind):

• Techno-charged authoritarians seizing power.
• “A swarm of millions or billions of fully automated armed drones.”
• “Sufficiently powerful AI could be used to compromise any computer system in the world.”
• Underestimating Putin.
• Silicon Oligarchy.
• AI will make war a lot easier.
• Powerful AI inventing a new religion.
• Virtual Bismarck!

According to Gemini 3, Virtual Bismarck registers a 7.5 on the Scary Scale. The “scary” part is not necessarily malice, but Virtual Bismarck’s competence. A Bismarckian AI would know exactly which “buttons” to push (economic incentives, social media algorithms, or diplomatic pressure) to get its way. Even if “its way” results in world peace, the idea that a black-box algorithm is pulling the strings of civilization is existentially chilling for most people. It is the difference between being a sovereign citizen and being a well-cared-for pet.

The chart above reminds us of the scary scale, with totalitarianism at the scariest end. Hannah Arendt would remind us that totalitarianism seeks total domination, aiming to control every aspect of daily life and eliminate human spontaneity. To be fair, Virtual Bismarck (7.5) is less scary than Silicon Oligarchy (8.5) or Corporate State (9). On the lighter side, Amodei’s explanation of the evolution of agriculture was pleasant.

Final Thoughts

If Machines of Loving Grace was the travel brochure for a post-AI world, The Adolescence of Technology is the survival manual for the trek there. Adolescence paints a picture of a future with “unimagined power,” provided we can survive the next few years of transition. While Amodei may be unaware of the subtleties of human nature, he is deeply worried about them. His belief that we can engineer our way out of human fallibility indicates that, while technology may be maturing, our leaders are still struggling with the complexities of the real world beyond the datacenter walls. It remains to be seen whether Amodei is a visionary guide or a sophisticated apologist for the inevitable. One thing is certain: the Cybertruck (adorned with a garish-colored local company’s logo) is already moving.