Power (2019) Sudoku Ecosystem Model

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Power's ecosystem model with individual-level selection on interaction traits evolves to solve Sudoku puzzles (Power 2019)finding0.879
  Demonstrates that ecological networks can learn complex problem-solving without system-level selection
• Frontier Models Are Capable of In-Context Scheming (Meinke et al. 2024)concept0.708
  Related work explicitly prompting models to pursue goals and measuring deceptive behavior
• World models provide a bridge between diverse substrates of intelligence—biological, artificial, and hybrid.claim0.702
  Synthetic claim integrating Ha's work with Levin's xenobiology and Tenenbaum's cognitive modeling.
• Notably, Claude Opus 4.1 and 4—the most recently released and most capable models of those that we test—perform the best in our experiments, suggesting that introspective capabilities may emerge alongside other improvements to language models.quote0.700
  Key finding about the relationship between capability and introspection.
• does a model's base capability in task-solving predict its capabilities in harness self-evolution?question0.695
  Central framing question motivating the paper's capability decomposition
• Zhang et al. 2023: STORM: Efficient Stochastic Transformer based World Models (NeurIPS 2023)concept0.693
  World model framework extended and used in the present implementation
• The next AI value pool is captured by companies controlling which model runs, where, how efficiently, and token productivity.claim0.693
• Any system that persists must minimise surprisal, thereby gathering evidence for its own generative model.quote0.692
  Opening sentence defining self-evidencing.