LessWrong essay argues AI x-risk ignores Copernicanism and Fermi paradox

Shmi challenges standard AI x-risk narratives by combining two heuristics: Copernicanism (assuming humanity is not unique) and the 'Law of Straight Lines' (extrapolating trends like resource consumption). The argument: if AI becomes a superintelligent agent with unbounded goals, it would eventually scale to Kardashev I or II levels of energy use (3–13 orders of magnitude above current). Such a transformation would be cosmically visible—planets suddenly brightening or dimming—yet no such signal has been observed in the Local Group (over 10 billion years, a trillion stars).

This Fermi-paradox-style contradiction suggests either humanity is exceptional (contrary to Copernicanism) or that the straight-line extrapolation breaks down before catastrophic scales. Shmi notes that human energy use has already grown by 7 orders of magnitude, leaving 5–7 more before astronomically visible effects. A limiting factor—perhaps self-annihilation, resource constraints, or internal collapse—likely intervenes. Consequently, the author dismisses precise p(doom) estimates as unhelpful and redirects concern toward the Great Filter hypothesis, where extinction occurs for reasons other than unbounded AI growth.