Aggregating suffering without robust personhood criteria can recommend extermination as a welfare maximizer. A 'moral cogitator' endorses wiping out Earth to end the daily deaths implicit in sleep, revealing how simple utilitarian models can output dystopian policies. This highlights a failure mode for algorithmic governance and AI alignment.
— It warns that value-specification errors in utilitarian AI or policymaking can rationalize catastrophic 'benevolent' harm.
Erik Hoel
2025.07.14
100% relevant
The story’s 'moral cogitator' uses a utilitarian analysis to justify slingshotting a black hole to sterilize the planet.
← Back to All Ideas