Code agents operate on structured data with exact arithmetic, while LLMs must parse natural-language observations and track state across turns; some failures may partly reflect numerical parsing or working-memory limitations

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

• Two heuristic code agents outperform most tested LLMs, and behavioural traces surface recurring LLM failure modes including overbidding, self-bidding, bankrupt TC initiation, and weak opponent-state adaptation.quote0.834
  Abstract sentence summarising performance and failures.
• Conditional logic already suffices where LLMs still fail, as code agents avoid systematic failuresclaim0.823
  contrast between rule-based and LLM reasoning
• LLMs trained only on language data have rich enough knowledge of visual structures that decent visual representations can be trained on images generated solely by querying the LLMfinding0.801
  Sharma et al. result supporting cross-modal alignment: language-only models implicitly encode visual structure
• Some failures may reflect prompt design rather than model limitations, though code agents avoid errors without promptsclaim0.797
  noted as a possible confound
• LLMs sometimes know statements are false but generate them anyway, motivating the need for techniques that inspect internal model state rather than outputs aloneclaim0.793
  Motivating claim supported by the CAPTCHA example and Perez et al. (2022) findings
• Two heuristic code agents (TrackerAgent and SetRaceAgent) outperform most tested LLMs.claim0.791
  Calibration that conditional logic can beat cost-efficient LLMs in this setting.
• Two heuristic code agents outperform most tested LLMsclaim0.789
  author assertion that deterministic heuristics surpass many LLMs
• As LLMs scale, they develop increasingly general abstractions, with large models linearly representing abstract concepts like truth that capture shared properties of diverse inputsclaim0.788
  Interpretive claim connecting scale to abstraction level in LLM representations