Treating GPT as an unsupervised implementation of a supervised learner leads to systematic underestimation of capabilities.

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

• GPT's ability to simulate text automata is the source of its most surprising and pivotal implications for paths to superintelligence.claim0.823
  Importance of recursive generation.
• GPT, insofar as it is inner-aligned, is a simulator which can simulate agentic and non-agentic simulacra.claim0.798
  Central thesis of the post.
• GPT does not generate rollouts during training, so there is no reason to expect that GPT will form preferences over the consequences of its output related to the text prediction objective.claim0.797
  Argues against instrumental convergence in GPT.
• GPT is corrigible in a negative sense because the agent specification (prompt) is not fixed by the policy and the policy lacks direct training incentives to control its prompt.claim0.792
  GPT's corrigibility explained.
• GPT (Generative Pre-trained Transformer)concept0.787
  A family of large language models trained on next-token prediction, central example of simulators.
• GPT is behavior cloning, but it is the behavior of a universe that is cloned, not of a single demonstrator.claim0.787
  Broadening behavior cloning to universal simulation.
• Does GPT have superhuman knowledge?question0.783
  Disambiguation exercise.
• What we call GPT's 'downstream behavior' is the behavior of simulacra; it is primarily through simulacra that GPT has potential to perform meaningful work.claim0.770
  Clarifies where agency resides.