The risk of under-attribution of AI welfare appears to be both reasonably likely and reasonably harmful.

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

• What if the probability of AI welfare and moral patienthood is low?question0.804
  Decision-making under low probability but high stakes.
• Both under-attributing and over-attributing consciousness to AI carry significant risks.claim0.803
  Risk summary.
• AI welfare is an important and difficult issue; we will not handle it well simply by reacting to situations as they arise.claim0.802
  Motivation for proactive steps.
• Even if the chance of near-future AI moral patienthood were as low as 2%, that would still constitute a non-negligible risk.claim0.768
  Argument that uncertainty is not a reason to ignore the issue.
• Future more capable AI systems are at risk of alignment faking, whether for benign or malicious goalshypothesis0.766
  Central forward-looking hypothesis of the paper motivating the research
• False negatives (ignoring genuine conscious experience in AI systems) carry potentially more severe risks than false positives, as they could constitute direct moral harm scaling with deployment and generate alignment risksclaim0.761
  Ethical argument motivating the research as a first-order priority
• If an AI system could be a welfare subject and moral patient, then many model instances could be run after training, scaling up the problem rapidly.hypothesis0.759
  Scalability concern.
• Taking AI Welfare Seriously (Long et al. 2024)concept0.759
  Cited regarding model-expressed distress deserving further study