instrument
◉ network⁘ semantic map⇄ tension compass▦ concept matrix
presets
the LLM debate the Levin cluster latent links talking past concept coverage
tension
T01 Do current LLMs possess a genuine world model / understanding, or only st… T02 How demanding must something be to count as a "world model" — any cognize… T03 Is embodiment / physical-temporal coupling with an environment necessary … T04 Is scaling the current autoregressive / next-token paradigm sufficient fo… T05 Does "emergence" in LLMs meet the scientific (complexity-science) standar… T06 Is genuine intelligence essentially compact and analogy-mediated ("less i… T07 Does behavioral performance license claims about internal understanding/w… T08 Should agency and selfhood be anchored in an enduring agent/self, or in d… T09 Are a world model's descriptive (state) and evaluative (reward/value) asp… T10 Is mind / world-modeling substrate-independent and continuous from single… T11 Is intelligence fundamentally collective, and is there a single generic m… T12 What best explains humans' greater sample-efficiency over current AI: str… T13 Is self-modeling a mechanistic self-regularizer (improving efficiency/pre… T14 Can a predictive / most-likely-next-step mechanism capture transformative… T15 Can current AI create genuinely new problems, concepts, or representation…
spotlight papers
○ Alicea et al.○ Amir et al.○ Battleday et al.○ Breithaupt○ Collins et al.○ Yiu et al.○ Hofstadter○ Krakauer et al.○ Ku et al.○ Leivada et al.○ Lyons et al.○ Northoff et al.○ Premakumar et al.○ Rouleau et al.○ Sacco et al.○ Safron et al.○ Tsividis et al.○ Zhu et al.
ask about this view → tension compass
Can current AI create genuinely new problems, concepts, or representational primitives (transformative respecification), or is it limited to solving pre-specified problems in a fixed frame? T15 genuine-disagreement
Current AI cannot autonomously create/respecify problems or primitives — it solves only pre-specified "easy problems"; this is (for now) uniquely human.
This is open or achievable — reflective respecification / self-pointed updating may be within reach, or primitives can be supplied by built-in core knowledge.
POLE A POLE B Battleday et al. Breithaupt Yiu et al. Tsividis et al. Collins et al. Hofstadter click a position for its verbatim quote