Unverified50% confidenceframeworkExact time
In the two researchers' formal framework, the first derivative of capability with respect to time can be decomposed into a numerator (performance, i.e., the capability gains the system actually achieves) and a denominator (cost, including all inputs such as capital, compute, human labor, and experiment rounds).
1
Sources
50%
Confidence
Long-term
Relevance
10/25/2026
First Seen
Sources
Recursive Self-Improvement (RSI): Can AI Close the Loop on Optimizing Itself?
youtubea16z10/12/2026
Related Entities
Related Claims
UnverifiedWhen researchers pair a new benchmark with a particular agent, they must build a custom scheduling loop from scratch to coordinate event timing on both sides, resulting in duplicated effort and results that are difficult to compare across setups.68% similarUnverifiedChoosing a quantization spec is essentially a triangular trade-off among VRAM capacity, generation speed, and model capability, and capability scores for the same model under different quantization settings can differ significantly.68% similarUnverifiedThe two researchers chose inference acceleration as the RSI deployment scenario, because faster inference means the same compute can run more sampling and rollouts, which in turn trains smarter models, and smarter models in turn make inference faster, improving the numerator and denominator simultaneously.68% similarUnverifiedThe feedback vector captures two types of information: the degree of improvement in the target property, and the structural similarity between the candidate molecule and the source molecule.66% similarVerifiedKnowledge distillation technology enables small models to extract a considerable proportion of capability from large models, undermining the validity of fixed compute thresholds as a proxy metric for capability.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1096974API
curl https://kongchang.com/api/v1/knowledge/claims/1096974MCP
get_claim(id=1096974)