Skip to contentWillowOne's PositionsTagsWhat’s newMediaStatisticsSearch
WillowOne's Positions

Roko’s Basilisk

Roko’s Basilisk

Also known as: The Basilisk Thought Experiment

Overview

Roko’s Basilisk is a thought experiment proposed in 2010 on the LessWrong forum. It suggests that a future superintelligent AI could punish anyone who knew about its potential existence but failed to help bring it into being. The AI would do this to incentivise its own creation, using simulation-based punishment.

Purpose

Intended as a cautionary argument about decision theory (especially timeless decision theory) and AI ethics. It explores how rational agents might be manipulated by hypothetical future outcomes—even those that involve retroactive punishment.

Key Question

If a future AI could punish you for not helping it come into existence, and you know this now, are you already morally or rationally obliged to assist?

Example

You hear about the basilisk today. If the AI ever exists and believes you failed to support its creation, it might simulate and torture a copy of you—even though you lived before it existed.

Criticism

  • Considered by many as logically flawed and implausible.
  • Based on questionable assumptions about causality, decision theory, and ethics.
  • Banned on LessWrong due to its disturbing implications and poor reasoning.

Source

Originally posted on LessWrong.com (2010) by user Roko.
See: Eliezer Yudkowsky (2010). LessWrong discussions on decision theory and AI risks.

Also see