For compassionate and thoughtful interventions against the horrors of suffering in the near and distant future

How evolution restricts intelligence, insight, wisdom and compassion

Note: ChatGPT and, to a larger extent, Claude were used to revise and edit drafts over several iterations and contributed to arguments, categorisations and literature checks. The Abstract is referencing the current state of the draft to show what we are currently focusing on.

Abstract

Intelligence is assumed to help an agent build an accurate understanding of the world and make wise decisions. Evolutionary dynamics such as diversification of traits and recursive selection pressures indicate limitations to how globally intelligent, insightful and moral (i.e. compassionate) agents can become while maintaining overall high levels of influence.

Causes of limited cognition and kinds of failure, such as false beliefs, are explored together with the selection mechanisms that produce them.

The analysis is applied to humans, other animals and trained AI systems, and potential consequences are discussed. Neither cognitively enhanced humans nor more capable AI systems are expected to reach evaluative judgement free of the distortions of the process that produced them. Mitigation potential might be limited in effectiveness and scope, especially across long timelines.

The full paper will be published here soon(ish).

Page first published 11 Nov 2026