entity · derived
Generative ai
Derived node: assembled mechanically from the claims carrying generative-ai. A roster, not an adjudicated definition.
Every claim under this term
- 4855607-015 : Reward modeling learned through interaction with users carries two structural pathologies: majority views disproportionately influence the learned reward function, and the agent may engage in reward h
- 5541658-010 : The Shenzhen courts illustrate a workable three step human-in-the-loop design for generative judicial AI: the judge decides first, the LLM generates the supporting reasoning, and the judge then revise
- 5541658-018 : Generative AI such as ChatGPT cannot produce justified beliefs aligned with virtue jurisprudence because it lacks the human virtues that responsive judging requires.
- 5541658-037 : Generative AI in judicial settings is far from perfect: it can oversimplify complex judicial deliberations, reduce emotive and cognitive processes to statistical correlations, and introduce biases or
- 5886442-004 : Generative AI unlocks 2.6 to 4.4 trillion dollars in annual value by automating 60 to 70 percent of work activities and reallocating 45 percent of working hours.