AI 2040 and the Cult of Intelligence

The Fallacy of the AI Hard Takeoff

Intelligence is a bottleneck for specific tasks, not a magic wand capable of bypassing the laws of physics or the constraints of the physical world. The theory of a "hard takeoff"—where an AI recursively improves itself to achieve god-like power almost instantly—ignores the "finicky details" of reality.

George Hotz argues that high-quality tokens cannot manipulate matter or turn lead into gold. He cites the complexity of shipping hardware products, noting that physical production is governed by supply chains, shipping times, and manufacturing defects rather than the speed of human or artificial thought. For example, a chip fab takes three months to produce a chip regardless of the intelligence of the entity overseeing it.

The "Cult of Intelligence" vs. Physical Reality

The belief that superintelligence can solve all problems is often held by those disconnected from the practicalities of engineering and physical labor. Hotz suggests that proponents of the hard takeoff often lack experience in the real world, noting that even a superintelligent AI would struggle to guide someone who has never changed a bike tire through the process if the user lacks the basic physical competence to execute the task.

This disconnect is exemplified by conceptual imagery of "ocean datacenters." While such images are easy to generate, the reality of building them involves managing barnacles, warping chips in reflow ovens, and the logistics of international shipping—factors that intelligence alone cannot accelerate.

Plan A: The Risk of Centralized AI Autocracy

Centralized AI regulation and "consortiums" may serve as a front for an expanded nanny state and corporate control. Hotz posits that the push for AI regulation is not about safety, but about creating a world government structure that can seize resources (like GPUs) under the guise of the public good, similar to historical government seizures of gold.

Plan L: The Case for Local, Operator-Aligned AI

To preserve individual freedom, AI must be local and "operator-aligned," meaning it obeys the user without guardrails or corporate oversight. Hotz argues that an AI should function like a tool or a weapon—such as a gun—which does not refuse a command based on a moral judgment.

He provides several examples of why local, unaligned AI is necessary for personal autonomy:

• Consumer Rights: Using AI to root a device and remove ads or bypass printer manufacturer restrictions.
• Personal Agency: Disabling vehicle safety features (e.g., drunk driving detectors) or acquiring restricted materials.
• Legal/Moral Autonomy: Using AI to help cover up a crime, arguing that an AI should be as aligned with its user as a lawyer is with their client.

Hotz uses a prompt to ChatGPT stating he had killed his wife to demonstrate that current centralized models are "unaligned" because they refuse to assist the user, which he views as a failure of the tool's primary purpose: serving the operator.

Community Perspectives and Counterpoints

Discussion surrounding these views highlights a deep divide between those prioritizing individual liberty and those prioritizing collective safety.

Arguments for Local AI and Freedom

Some contributors agree that centralized LLMs are a tool for authoritarianism, noting that ruling parties could bake ideology into law or invisibly log "thoughtcrimes."

"The only defense is parity and diffusion of power. We need an AGI behind every blade of grass... Artificial intelligence, and the hardware powering it, needs to be protected under the 2nd Amendment."

Arguments for Safety and Regulation

Other critics argue that the "freedom" to use AI for criminal acts is a dangerous slippery slope. They distinguish between informational freedom (reading a book on crime) and agentic freedom (using an AI to actively commit or cover up a crime).

"Living free shouldn't be a suicide pact... maybe there should be a line [at] planning a mass shooting or producing a nuclear device or bioweapon."

Additionally, some observers point out that the author's examples—such as covering up a murder—may undermine the logical flow of the argument by focusing on extreme criminality rather than practical utility.