Gordon Seidoh Worley's post on AI Alignment Forum and LessWrong argues that achieving AI alignment inherently requires some form of dependence. The author suggests that true alignment cannot be fully endogenous, meaning it cannot arise solely from within the AI system itself without external factors or relationships. This perspective challenges purely internal approaches to AI safety and control. AI
IMPACT Challenges purely internal approaches to AI safety, suggesting external factors are necessary for alignment.
RANK_REASON The cluster consists of two posts by the same author on different platforms discussing a theoretical concept in AI safety, fitting the definition of commentary.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
