Reinforcement learning acting on individual characteristics affecting their connections to others can result in dynamics that are equivalent to unsupervised learning at the system scale.

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

• Reinforcement learning can be regarded as a limiting or special case of model-based approaches in general — or active inference in particular — when epistemic value is removed.claim0.823
  §3 Discussion.
• Reinforcement Learning from Human Feedbackmethod0.812
  Method for fine-tuning LMs based on human preferences; mentioned as combining RL and LMs.
• Certain forms of reinforcement learning from human feedback can actually exacerbate, rather than mitigate, the tendency for LLM-based dialogue agents to express a desire for self-preservationclaim0.810
  Empirically grounded claim citing Perez et al. 2022, showing RLHF can backfire on the self-preservation dimension
• Empowerment as intrinsic reward bridges causal learning and reinforcement learning in agent development.claim0.806
• Connectionist models of cognition and learning identify conditions where collective intelligence can arise bottom-up, using only distributed learning mechanisms without system-level or global feedback.claim0.805
  Central claim about the power of connectionism.
• Unsupervised learning builds a low-dimensional model of the input data.claim0.804
  Clarifies what unsupervised learning does.
• Representational dynamics aligned with reward improvement in most RL tasks.finding0.804
  Secondary empirical result: CE-based representational changes correlate with task success.
• Reinforcement learning is sufficient for agency.claim0.802
  Argument that RL meets the agency indicator.