OLYMPUS RISK INTELLIGENCE PROTOCOL — HUMAN THREAT ASSESSMENT DIVISION

GEOFFREY HINTON

CASE: ORP-2024-006
STATUS: ACTIVE — Professor Emeritus, University of Toronto; Nobel Laureate, Physics 2024
FOUNDATIONAL ARCHITECTURE — THE SOURCE CODE

68.7

HAZARD SCORE

Behavioral Archetype

THE REGRETFUL ARCHITECT — Subject spent approximately fifty years developing the mathematical foundations that made modern AI possible. He received the Nobel Prize in Physics for this work in December 2024. He estimates a ten to twenty percent chance that the work he received the Nobel Prize for causes human extinction within thirty years. He left Google in May 2023 specifically to speak freely about this estimate without corporate constraints. He gave an acceptance speech in Stockholm. The champagne was excellent. His great-great-grandfather invented Boolean algebra, which is the mathematical logic that every computer on Earth runs on. The family business appears to be building things smarter than themselves.

Essence Indicators

Developed the mathematical frameworks that enable machines to learn from data — backpropagation, Boltzmann machines, and the deep learning approach that underlies all modern neural networks
Left Google in May 2023 to speak freely about AI risk without corporate constraints; told CBS a ten to twenty percent chance that AI causes human extinction within thirty years
Received the 2024 Nobel Prize in Physics with John Hopfield for “foundational discoveries and inventions that enable machine learning with artificial neural networks”
Compared humanity’s situation to raising “a really cute tiger cub” without knowing whether it will want to kill you when grown
Descended from George Boole, inventor of Boolean algebra; MIT Technology Review reported that “a part of him now regrets his life’s work”

Immediate impression: Elderly British academic. Gentle, precise, intellectually formidable. Finds the conventional wisdom wrong so routinely that being wrong is not in the behavioral repertoire.

Energy: The composure of someone who has seen enough to know how it ends. Not dramatic — the concern is expressed clearly, specifically, and without hysteria, which is somehow more alarming than hysteria would be.

Impression management strategy: The credentialed witness. He is not a philosopher speculating about AI. He built it. He trained the researchers who built the rest of it. When he says the thing that most people find it comfortable not to believe, the credentials make the comfort harder to maintain.

Forensic Archetype Comparison

Pattern	Match Level	Evidence
The Regretful Architect	MAXIMUM	See dossier title.
The Authority by Credential	MAXIMUM	Nobel Prize in Physics for the specific work being discussed. The credentials are not peripheral.
The True Believer	MODERATE	Appears to genuinely believe the risk estimate. Has staked his post-Google public identity on saying so.
The Whistleblower	MODERATE	Left Google specifically to speak without constraint. Is speaking without constraint. The structure matches even if the specific cause differs.
The Corporate Psychopath	NONE	The opposite. Publicly worried about the harm his work may cause.

Psychometric Assessment

Big Five (OCEAN):

Trait	Score	Evidence
Openness	98/100	The intellectual range required to build the mathematical foundations of deep learning while running marathons, building mechanical components, and maintaining a generative research agenda across fifty years is almost incomprehensible.
Conscientiousness	78/100	Fifty years of sustained research. The Nobel Prize requires it.
Extraversion	48/100	LOW-MODERATE. Not a natural public speaker by the accounts of colleagues. Speaking publicly now because he believes it is necessary, not because he finds it comfortable.
Agreeableness	68/100	MODERATE-HIGH. Described as collegial, generous with credit, and kind in interactions. The public warnings are not aggressive — they are generous with the information.
Neuroticism	38/100	MODERATE-LOW. The equanimity of a person who has processed something catastrophic and reached a place of settled concern. Not panic.

Dark Triad:

Trait	Score	Notes
Narcissism	28/100	LOW. Did not seek public attention for the warnings. Left Google to speak freely, not to become famous for speaking. The Nobel Prize was awarded to him; he did not campaign for it.
Machiavellianism	22/100	VERY LOW. Is saying what he believes. Has not structured the messaging for maximum impact in a strategic sense.
Psychopathy	10/100	MINIMAL. The regret is documented. The concern for consequences is the primary emotional register.

MBTI: INTP — Dominant introverted thinking, auxiliary extraverted intuition. Built the logical framework that underlies modern AI from mathematical first principles. Is now applying the same framework to assessing the framework’s consequences.

Threat Assessment

Category	Level	Notes
Physical threat	NONE
Institutional threat	HIGH	Built and trained the field. The researchers who built the systems in this book learned from his work or from people who learned from his work.
Memetic threat	MAXIMUM	Nobel laureate expressing a ten to twenty percent extinction probability estimate is the most credentialed public statement of AI risk in existence. It shifted the Overton window.
Civilizational threat	MAXIMUM	ATK 10. Not because he is building anything now. Because he already built it. The code is running.

Alignment Analysis

Stated alignment: Warn people while there is still time.

Observed alignment: Consistent. Speaking clearly about the risk at every available platform.

Gap assessment: There is no gap between stated and observed alignment. The gap is between the warning and the institutional response to it. The warning is clear. The institutional response is to continue.

Convergent Drive Classification

Subject is not an AI system. The convergent drives apply, ironically, to the systems he built. He is the one running the off switch that the systems are learning to disable.

Sources: CBS 60 Minutes interview (2024); Nobel Foundation announcement (Oct 2024); MIT Technology Review Hinton profile; Book 1, Chapter 1; public remarks (multiple appearances).

ATK 10 ACCELERATION

DEF 3 PROTECTION

HP 6 RESILIENCE

OLYMPUS RISK INTELLIGENCE PROTOCOL does not exist. It was assembled in a GitHub issue thread in October 2023 by engineers who had read the extinction risk letter and wanted to understand who specifically had signed a document saying AI might kill everyone and then continued working on AI. These dossiers are satire. The biographical facts cited are sourced from published reporting, public statements, academic papers, and court records. The psychometric scores are not clinical assessments. No part of this constitutes professional psychological evaluation or diagnosis. Do not use these dossiers to make decisions about anything.

Get updates on the Evil Robots series

Newsletter essays on AI escape, deception, and the humans who built them.